CLAUDE COWORK MASTERCLASS for Professionals+Claude AI No Code | Paul Ashun | Skillshare

Playback Speed


1.0x


  • 0.5x
  • 0.75x
  • 1x (Normal)
  • 1.25x
  • 1.5x
  • 1.75x
  • 2x

CLAUDE COWORK MASTERCLASS for Professionals+Claude AI No Code

teacher avatar Paul Ashun, Deliver Projects On Time with AI Agile & Scrum

Watch this class and thousands more

Get unlimited access to every class
Taught by industry leaders & working professionals
Topics include illustration, design, photography, and more

Watch this class and thousands more

Get unlimited access to every class
Taught by industry leaders & working professionals
Topics include illustration, design, photography, and more

Lessons in This Class

    • 1.

      Introduction

      3:59

    • 2.

      The Birth of AI and LLMs

      1:43

    • 3.

      What Is an AI Agent (And Why Cowork Is One)

      7:40

    • 4.

      Which Plan Do You Need For Cowork

      8:22

    • 5.

      How To Install The Claude Desktop App And Set Up Cowork

      6:34

    • 6.

      How To Grant Folder Access So Claude Can Read, Edit, And Create Files

      3:30

    • 7.

      How To Use Projects To Organise Your Work And Build Persistent Workspaces

      11:34

    • 8.

      How To Set Permissions And Stay In Control Of What Claude Can Touch

      8:20

    • 9.

      How To Assign A Task And Let Cowork Handle The Steps

      7:29

    • 10.

      How To Organise Messy Folders Using Cowork

      10:43

    • 11.

      How To Turn Raw Data And Screenshots Into Finished Spreadsheets

      10:24

    • 12.

      How To Generate Reports And Documents From Your Files

      15:01

    • 13.

      How To Schedule Tasks So Claude Runs Work On A Set Cadence

      11:44

    • 14.

      How To Set Up Recurring Workflows And Walk Away

      15:22

    • 15.

      How To Use Claude In Chrome To Browse And Research On Your Behalf

      15:14

    • 16.

      How To Use Connectors To Link Cowork To External Services

      20:50

    • 17.

      How To Use Skills To Give Claude Domain Knowledge And Best Practices

      13:42

    • 18.

      How To Extend Cowork Using Plugins From The Marketplace — And Build Your Own

      9:35

    • 19.

      How To Use Dispatch To Assign Desktop Tasks From Your Phone

      14:33

  • --
  • Beginner level
  • Intermediate level
  • Advanced level
  • All levels

Community Generated

The level is determined by a majority opinion of students who have reviewed this class. The teacher's recommendation is shown until at least 5 student responses are collected.

9

Students

--

Projects

About This Class

In this course, you'll learn how to move beyond AI chat and simple prompts to use Claude Cowork , Anthropic's agentic AI platform  to execute real tasks, automate your workflows, and get finished work done without writing a single line of code.

You'll go from understanding what agentic AI actually means to running intelligent, multi-step workflows that connect to your files, your apps, and your data — and deliver real outputs you can use immediately.

We'll start with the foundations of Claude Cowork and why it's fundamentally different from the Claude you may already be using. Then you'll learn how to give Cowork access to your work, use Projects to build persistent memory and context, and assign tasks that Claude executes end-to-end — reading files, making decisions, and saving finished outputs — while you stay in control.

A core focus of this course is building workflows that actually run your work for you.

You'll learn how to:

  • Organise messy folders, process data, and generate polished reports from raw files

  • Schedule recurring tasks so Cowork runs your regular work automatically — daily, weekly, or on any cadence you set

  • Connect Cowork to Gmail, Google Drive, Google Calendar, Microsoft 365, Slack, and 38+ other apps

  • Build multi-app workflows that span connectors and local files in a single brief

  • Use Skills and Plugins to give Claude deep domain knowledge for your specific role

From there, you'll go further with advanced Cowork features, including:

  • Live dashboards and trackers that auto-refresh from your data

  • Computer Use — letting Cowork control your screen for tasks with no connector

  • Dispatch and mobile pairing — assigning tasks from your phone and coming back to finished work

  • Sub-agents that handle specialist tasks end-to-end with built-in domain expertise

You'll also master practical, real-world applications of Cowork for:

  • Document creation, report writing, and data extraction from PDFs and contracts

  • Customer feedback analysis and supplier management

  • Marketing content, social posts, and product launch workflows

  • Multi-app automation across your entire tool stack

  • Recurring business operations — running automatically, without you triggering them

By the end of this course, you won't just be using AI.

You'll have an intelligent, automated operation running around your work — one that reads your files, executes your workflows, connects to your tools, and delivers finished outputs while you focus on the decisions only you can make.

This is a hands-on course built around live demos, real workflows, and a running business scenario used across every lesson — so you can see exactly where to click, what to type, and how to apply it to your own work immediately.

No coding. No technical background required. Just real work, done faster.

Meet Your Teacher

Teacher Profile Image

Paul Ashun

Deliver Projects On Time with AI Agile & Scrum

Teacher

What do students say?

"I liked the course. It was quick and easy to understand, but also complete. Thank you."

"The course gets to the point. Great course, it's short and show all the points to get the scrum certification."

"Excellent Material!Thanks for the clear cut training material."

I am grateful to have received this feedback from a fan because it explains exactly the value I hope to give you in my courses!

► Enroll in one of my courses today to save hundreds of hours learning the hard way and thousands of dollars on training courses like I did! ◄

What qualifies me to share my experience with you?

1. I can help! I am a Scrum expert and have lead projects as a software engineer, tech lead, team lead, scrum master, program... See full profile

Level: All Levels

Class Ratings

Expectations Met?
    Exceeded!
  • 0%
  • Yes
  • 0%
  • Somewhat
  • 0%
  • Not really
  • 0%

Why Join Skillshare?

Take award-winning Skillshare Original Classes

Each class has short lessons, hands-on projects

Your membership supports Skillshare teachers

Learn From Anywhere

Take classes on the go with the Skillshare app. Stream or download to watch on the plane, the subway, or wherever you learn best.

Transcripts

1. Introduction: Thank you, and congratulations on taking this course, Claude Cowork Masterclass, how to use agentic AI to automate your work run intelligent workflows and get more done. Without writing a single line of code, if you've ever used Claude or any AI tool, type something in, got a great answer, and then thought, How can I get this done without repeating myself? How can I save more time and get the job done way faster? How can I get AI to do the task instead of giving you the answers? You're in exactly the right place because we've entered the new era of AI, not AI that responds AI that acts autonomously. On your machine with your files across your apps. This is what's called agentic AI, and Cowork is anthropics implementation of it. Built specifically for professionals who want to get real work done without writing a single line of code. Let's start with a real scenario. You're running a business, managing a team or working in any busy professional role. Every week, you're doing the same things, generating reports from messy spreadsheets, processing orders, reading, and responding to emails, preparing documents, Word files, PDFs, forms you fill in the same way every time. Chasing supplier invoices, sitting in meetings that could have been a summary, and every week that work takes hours, hours that could be spent on the decisions only you can make, the strategic thinking, the relationships, the work that actually moves things forward. Here's what changes with cows like having an intelligent assistant working for you around the clock, one that never gets tired, never complains, and never needs things explained twice. Whether you're a business owner, a manager or an assistant yourself, that kind of support changes what you're capable of. And in this master class, we're going to show you exactly how to set it up, step by step, demo by demo, using a live running business scenario throughout every lesson. Here's what you'll learn. You'll start by understanding what cow actually is and why it's fundamentally different from the chat version of Claude you may already be using you'll learn how to use projects. So Claude builds memory and context about your work across every session. You stop explaining, it starts knowing. You'll learn how to give Cowork access to your files and folders, so it can read, create, edit, and organize on your behalf. You'll learn how to assign tasks and let Cowork handle all the steps, including how to review its plan before it acts and how to step in and redirect it in real time. You'll learn how to work with documents turning raw data and screenshots into finished spreadsheets, generating polished reports, and extracting structured information from dense contracts and PDFs automatically. You'll learn how to schedule tasks, so Cowork runs your regular work on a set cadence without you triggering it. You'll learn how to connect Cowork to your apps, Gmail, Google Drive, Google Calendar, Microsoft 365, Slack and dozens more, and build workflows that span all of them in a single brief. You'll learn how to use Skills and plugins. So Cowork arrives at every task with deep domain knowledge already built in. And you'll learn how to use dispatcher, so you can assign tasks from your phone and come back to finished work, even when you're nowhere near your desk. By the end of this course, you won't just be using cow. You'll be running an intelligent automated operation around your work, one that reads your files, executes your workflows, connects to your tools, and delivers finished outputs, while you focus on the things that actually need you. Every lesson is practical. Every demo was created live. Every example is built around a real business scenario you can follow along with and apply to your own work immediately. Take your time, follow the demos, experiment as you go. And by the end of this course, you'll be operating at a level most people around you simply haven't reached yet. So let's get started. 2. The Birth of AI and LLMs: AI, the birth of chat, GPT, and LLMs. From the simple rule based programs of the 1980s to today's smart creative chatbots, one thing has stayed the same. Our goal to make computers understand and respond similar to humans. The story of chat, GPT, and large language models, LLMs, is really the story of how that dream became real. How we went from basic machines to powerful tools that can think and write using everyday language. Although it feels new, artificial intelligence AI has actually been around for decades. To understand where Chat GPT came from, let's take a quick look at how AI evolved over time. It all began in the 1950s when computer scientist Alan Turing asked a famous question. Can machines think? That one question started the entire field of AI. In the 1960s, a simple program called Eliza was created at MIT. Would hold short conversations by matching patterns in text. It wasn't truly intelligent, but it was the first step toward computers that could use language to communicate. In the 1980s, AI was used mostly for what were called expert systems, programs that followed rules written by humans. These systems could give medical advice, approved loans, or help design products. They were useful, but they had one big problem. They couldn't learn or adapt. If something changed, you had to rewrite the rules yourself. The 1990s and 2000 brought a big shift, the rise of machine learning. I 3. What Is an AI Agent (And Why Cowork Is One): So in this lesson, we're going to talk about what an AI agent is and why CW is an AI agent. So before we even install anything, we need to understand the shift that makes cow different from every other AI tool that you've used before. And in this lesson, we cover AI agents, what they are, how they think, and why CW is the most accessible version of this technology for knowledge workers or for anyone else. So let's get started. So the feature we're covering is AI agents and co work. Why do we use them? We use them to understand the shift from generating text to executing real tasks autonomously. And what it solves, it solves using cowork like a chat bot and getting 10% of the value, because if you start using cow the wrong way, you're not really going to see the value. So it's important to understand what an AI agent is, how we use it, and how cow fits into this to help you save time and be more productive. So first of all, first of all, let's talk about generative AI. When we use something like clawed chat, we're using generative AI. And the way it responds is you type a prompt, and then the AI generates a reply. So one input, it generates one output, and then it stops. It doesn't do anything next until you tell it to. Now, with an AI agent, also known as agentic AI as opposed to generative AI, be using something like clawed CW, what we're about to use. Now, an AI agent both plans and executes. So you define a goal, the agent reasons first, then it plans a sequence of actions, uses tools, and it executes, and it executes autonomously until the work is done. And we're going to break down what autonomously means. So here's a little comparison table about generative AI versus agentic AI or in other words, generative AI with a chatbot versus an AI agent like Claude Cork. So generative AI is something like clawed chat, what you're used to using with clawed chat, the usual what most people know as clawed. And what it does is it responds to your message, whereas clawed cow plans and executes towards your goal. In terms of file access, clawed chat accesses files only if you upload, whereas clawed cow has direct access to your local files and apps. In terms of autonomy, Claude Chat has none, so it always waits for your next message, whereas Claude cow runs independently, and it loops you in as needed. So it may stop and say, Hey, are you happy for me to move forward? Whereas Claude Chat will just do exactly what you tell it to do and then wait for your next message. And even then it can only really grab information for you except for a few exceptions. In terms of what they're best for, Claude Chat is best for writing, Q&A, drafting, and analysis, whereas Claued CW is best for organizing, reporting, automating, and executing. Now let's go through a few key concepts that we really need to understand in order to understand AI agents and clawed cowork. So the first thing is autonomy. Autonomy means an agent can work without being directed at every single step. You set the goal, it figures out the sequence of actions needed to reach that goal. How does Cork do it? Cork shows you the plan first, you approve, and then it runs every single step without interruption. You can step away and come back to completely finished work. So that's the big difference here. The next concept is a concept of reasoning. What reasoning means is that agents don't follow instructions literally. They reason about what needs to happen, they think before they act, and this is one of the powerful things about the agentic era, AI agents. So, for example, if you ask cow to organize the supplier files, it looks at what's there, what files it has access to. It infers different categories maybe for those files. It will propose a structure, and then it will apply the structure to those files. And this is without you specifying every single rule and how it should apply to structure. It can make some decisions for itself by reasoning. Next concept is a concept of tool use. So what that means is that agents use tools, and that could be file systems, web browsers, app connectors, or different apps to get things done. The agent picks the right tool for each step automatically, and that's what makes the agent powerful, but it has access to these tools, and it picks them itself. And then CoWors tools that you can use are local folders, the Chrome browser, about 38 plus app connectors and growing, such as Gmail, Drive, Slack, Notion, Microsoft 365, and many more. And then Cowork selects the right one for each step, and you don't even have to ch. There's a level of autonomy there as well. But the main thing here is that as an AI agent, the difference between the AI agent and the previous Claude chat and even chat GPT is the Claude cow agent can choose the right tool for each step. You don't need to choose at all. The next concept is memory and goal directed behavior. So in terms of memory, agents retain context across a session. So CWT remembers what it's working on, what files it's already seen, and what decisions were made, so you don't have to repeat yourself because it's all in memory. And then goal directed behavior means that unlike a chat bot like normal ClaueI that responds to the last message, an agent stays focused on the original goal. And that's even across many steps, it's still focused on the original goal. So CoWor will keep asking, Am I closer to the outcome that the user defined, and it will ask itself that question, and it will only come and ask you a question when it really needs your input. For the most part, it will just carry on autonomously, making decisions and moving towards your goal until it reaches your goal. And it will remember that, too. So if we break it down into two different approaches, the chat approach, which is generative AI and the agentic approach, the cow approach, which is agentic AI. So let's take the goal to turn a folder of order files into a weekly sales report. With the clawed chat approach, we would, number one, upload each order CSV file manually. Two, we would paste the data or describe it in a message. Three, Claude writes a report from what you gave it. Four, you copy the output, format it, and save it yourself. And this would take about 20 minutes, four manual steps, as you can see, and the output isn't actually saved anywhere unless you save it. Now, with the cow approach, it's the same goal, but you write one brief. So let's say you type in the prompt, read Passion Sports orders Week 18 and produce a weekly sales report, save to Passion Sports slash report. So you're reading from the orders in one folder, and you're telling cow to produce a weekly sales report and save it to another folder. Let's say that was the goal, and we wrote that all in one brief, all in one prompt. What cowork does from here is number one, it will show its plan, you approve its plan, and then from this point, cow takes over. It reads all the order files in a folder, analyzes the data, and writes the report all on its own, saves the finished file directly to your reports folder, and all of this can happen in 30 seconds, sometimes less from one brief, all the files saved automatically, and you didn't have to do anything. No uploading or downloading or even thinking about it. Claude cowork reasons and does it all itself autonomously. And this is just one really simple example. There are thousands in summary, generative AI responds to prompts. AI agents plan, reason, and execute towards a goal. CoW is an AI agent. It works autonomously across multiple steps to reach the outcome you define. Key agent concepts are autonomy, reasoning, to use, memory, and goal directed behavior. CoW is uniquely desktop native, direct file access that cloud agents can't match, no coding required, and you stay in control. CoW shows its plan and loops you in before significant actions. All you have to do is just approve it, and CW does all the work for. Now it's your turn. It's time to work out your first agent task. What I want you to do is think of one task you do every week in your own business or your role that involves multiple steps, such as collecting data, processing data, writing a summary, anything at all. And then write it down as an outcome and think to yourself, what does finished look like for this task? What files are involved, and what is the output? This is your first candidate for a cow agent task, and we're going to build it later on in this course. So have fun doing that, and I'll see you in the next lesson. 4. Which Plan Do You Need For Cowork: In this lesson, we're going to cover which plan you need for cow. So what we're doing today, we're going over the fact that CW isn't available on every Cloud plan. And before you install anything, you really need to know which plan unlocks the features in CW and that's so that you don't hit a brick wall in the middle of the lesson. And the thing that I want you to know is that plans change regularly, pricing plans, that is. So always check claw.com slash PRICE for the latest or go to claw.com and look for the pricing page. The information that I'm about to go through is correct as of today, and it still can change. So let's get into it. So what we're covering today is clawed cow access and how we access it. Why we're using this why we're looking at this particular feature is to unlock the agentic desktop agent. So you download the clawed desktop app. But once you're in there, the only way to unlock it is with the right plan. And what it solves is the issue where you don't know which plan you're going to need before you start. So basically, we're saving you from having any wasted time. So let's go over the plan. So the first one here is the free plan, zero pounds a month or $0 a month. And basically, with that, you can't get access to work. You must have a paid plan. So this is best if you're exploring clawed chat, which I've done in another training, which is the basic chat bot that allows you to ask questions and get feedback. It also does allow you to do things like create artifacts, which are things like documents, charts, and things like that. So it is very helpful, but it won't give you access to cow. The next is the pro plan, and that's the minimum that you need for this course. And you can use cow in full with all features, and it's good for daily professional use. Next, we've got the MAX plan, and that gives you full cow access, but the only difference is it gives you higher usage. And because it gives you higher usage, you can get more done and we'll go over that a little bit more in the demos, you'll see. But to frame that slightly, it's best for power users who are running cow all day. So let's say you've got a number of tasks, and those tasks are things you might do once or twice a day. You'll probably get away with it with professional use. But if you're running tasks throughout the day, then you're likely to need the MAX plan, and I'd consider you a power user because you'll be using cowork all day. And these things will become clearer as we go through the training, a little bit difficult to say right now because differ People, depending on the role that you're in or the work that you're doing, use cow to different extents, so it will become clearer throughout the training. Then next, we have the team plan, and per seat is essentially per user. And again, that gives you full cow access, as the ones above do, but it also gives you admin control, so you've got more control over how you administer with the team plan. That's not something that I'm going to be going through, so we're sticking with a pro plan. And so this is the reason why you need that admin those admin rights are for teams deploying co work together because often what you have is teams who need to interact with each other. And therefore, that's why you've got the concepts of seats for each user or team. Then the next plan is enterprise, and that is custom pricing, meaning that you need to reach out to Anthropic who built Claude, and they will determine your price based on the usage, needs, and the number of seats, and things like that. And that is with that, you get full coor access. And that means role based access control. And this means security information and event management. What that basically means is role based access control means you might have, say, different levels of access to co work for the financial controller in your company versus a project manager in your company. One needs to see all of the financial information, all of the financial tasks that you're dealing with and doesn't want to see all the project management ones, whereas a project manager would want to see all the project management tasks, but wouldn't want to see all the financial ones. And for security reasons, you want to partition that. And SAA allows people to feed Claude's activity logs into their security monitoring so they can monitor what's actually happening within Claude. So big enterprises would really benefit from that. And so that's best for large organizations. So we're not going to that level. Pro is perfectly fine to move forward with this training. So the minimum requirement for this training is pro plan or above, and this course requires that, and everything in the outline, the task execution, plug in, skills, dispatch, all those features, computer use, live dashboards can be done safely with Pro. If you're on the free plan, you won't have access to a cowork at all, so you can go to upgrade at clade AI slash Upgrade or go to clad.com and search for the ability to upgrade all within Claude. So here we are at claw.ai, and I've gone to that within Chrome. I'm not using any downloaded apps, so this is something you can do as well. It will get you to log in. But once you're logged in, you can upgrade your plan by going down here and then going to Upgrade Plan. So if we click on that, you'll see you'll have the options here to upgrade. And what you want is the pro plan here. So the pro plan will give you the price, which will, as I say, be around 15 to $20 a month in the US and the same in the UK roughly. So then you can just choose either to have annual plan or to have a monthly. I'm already on the monthly, so it's only offering me the annual plan. And that way, once you're upgraded, you'll be able to use Claude cowork which we haven't installed yet, but I will show you in another lesson. And remember, this is the individual set of plans. So what you can do, also, if you're doing this for your company, you can get the team plans here or the enterprise plans. So you can just choose what's best for you. But as I've said, all you need is the pro plan for this. So let's talk about usage because one thing particularly running clawed cow is depending on the amount of tasks that you are going to run, you will use potentially more processing power and more compute, which is going to affect the plan that you want to pick. So the way Claude usage works is Claude doesn't give you a fixed number of tasks per day or anything like that. Usage is based on your session. So heavy tasks where you're doing things like reading many files, running long workflows that consume more time consume more of your session, simple as that. So what you need to understand is between the different plans that you pick that Pro will give you five times more usage than the free plan. You can't even use the free plan, but if you've been using Claude AI to chat, then you will actually see the amount of usage. That you've used so far, and it'll give you a good idea. And you'll know that the amount that you use on the free plan takes you to a certain level before you're out of your usage for that period of time. So thinking about that, you know that you'll get five times more usage than the free plan should give you an idea of how much usage you'll have within cow even though the tasks are likely to be more intensive. And then the MAX plan gives you five to 20 times more usage than the pro plan. So for users who are running cow all day across many files, that will give you some idea of what you'll use. And again, you'll know more about it as you start to go through the task in the training. But the key thing to know is that the usage resets periodically. So what you want to do is go to claw.com slash PRCI for current limix or limits, or just go to claw.com and look for the pricing, and then make your decision based on that about which plan you go for. So in summary, CW requires either the Pro Max team or enterprise plan at this time. The free plan has no CW access. The pro gives you everything in this course, and the usage is session based. It's not a fixed task count per day, so do X amount of task, and then you run out of your limit. It's session based. So within your session, for a fixed amount of time, you can use as much as your session allows you on the P plan, and Claude Cook will tell you when you're coming near the end or at the end of what you can use in your session, and then you need to wait for it to refresh, which is usually every 5 hours, there's weekly limits as well. But Claude will tell you when you're out of session, and then you can choose to upgrade. And check, as I said, usually cloud.com slash PRICE for the latest orgcla.com and check the pricing on there and know that the plans update regularly. So just keep in touch with that as time changes, and you will always know where you stand with pricing. So there you go. And now it's your turn. What I want you to do is check your plan. So what I want you to do is check your current Claude plan at claude AI. Or look at the bottom left of the sidebar and go to your upgrade or your current plan, the way I've shown you. And if you're currently on the free plan, then what I want you to is choose P before the next lesson. If you're on Pro or above, then you're ready to go. You don't need to do anything, and you can move to the next lesson. So have fun doing that and I will see you in the next lesson. 5. How To Install The Claude Desktop App And Set Up Cowork: So in this lesson, we're going to go over how to install the clawed desktop app and set up cowork. Now, CW lives inside the clawed desktop app, not the browser, so you can't use cow on claw.ai on your web browser. So this lesson gets you from zero to your first cow session ready to run. So let's get started. So the important thing here to remember is that cowork is desktop only. It runs on your MAC OS and Windows X 64. They're both supported as of April 2026. And Intel Max and Windows ARM 64 are not supported. So the desktop must stay open while Co Work runs, and closing it ends the session. But remember, on these particular platforms, Intel Max and Windows ARM 64 is currently not supported. That may change by the time of this lesson, but currently not supported. CoWor downloads a Linux virtual machine, which is around two gig on first launch at the moment, and so therefore allow time for this download. So here we are inside of the Chrome browser, and we're at cloud.com slash Download. And you can also find it by going to Cloud AI and looking for the download button. Either WillWO. And what I see because I'm currently on a MAC is it shows me the download from Mac OS Ston because that's what will allow you to use the full feature set with CWC. So I'm going to click the Download from Mac OS Ston and that's going to download the file that we need the DMG file, which is the Install file on a MAC And that is going to download over here on Chrome, and we're just going to wait for that to download. So now that that's downloaded, I'm going to open up the folder it's in and click on the DMG file to run it, and that will install Claude desktop. And now that that's downloaded, you can see here that there's the clawed icon. And so what I'm going to do is drag that into the applications folder because that's where all my applications exist. And you see now it's loaded into the application folder. So if I click into the application folder, you can see here are all the applications. That's Claude. So now we need to do really is double click on Claude and it will open up immediately. You see that it asks it says that Claude is an downloaded from the Internet. Are you sure you want to open it? I believe I'm sure about that, so I'm going to open it. So now we've opened up the clawed desktop application. You can see for me that there are a few chats down the bottom here because I've actually installed the application before and I deleted it so that I can show you how to install it. So for you, this area should be completely clear. And you'll be familiar with the chat box, which is what you would see on clawed.ai, if you've used that before, or Chat TPT, very similar. But the important thing is on the left hand side, you'll see that there's a left menu, which I'll go through at a later date. And above it, you can see that there are three tabs. You've got the chat tab, which is what you would be familiar with if you've used Cloud AI before. And on the right hand side, you've got the clawed code option, which is, if you're a developer and you want to write some code, but what we're going to be using is we're going to be using clawed cowork which is this tab here in the middle. So make sure that that is the one that you selected first. And then we can come over to the chat box where we're going to paste in our prompt and our instructions. Now, what we want to do is we want to make sure we've got access to the file system. And usually, as you can see down here, we can select which folder we're going to work in. But before we do that, we're going to run a little test, and we're going to do that by pasting our prompt like so. So our prompt says, create a text file in my documents folder called CW underscore text test dot TXT. Containing the words cow is working correctly. So that's a little test to make sure that we can access our file system. So let's run that. So you can see we'll authenticate, first of all, says it's working on it. And then eventually, after a little bit of progress, it says, I don't currently have access to your documents folder. Let me request it so I can create the file there. So this is standard for Claude. You'll notice that it will check before it's about to do anything sensitive. Now, we just need to decide whether we want it to go to the next step, so I'm going to allow it, click the allow button. So it continues working on it, and it now says, Got access now I'll create the file. We see it's writing the file now. And there you go. It says Done. The CW underscore test dottXT has been created in your documents folder with the text CWC is working correctly. So let's go over and have a look at that file and make sure it's created it correctly, as it said. So I'm going to go to myFinder so that I can find the file. And I'm going to look inside of my documents folder. And there is CW Underscore test dotXT. So let's open that. And there you go. It's got some text inside it. So if we zoom in a little bit, we can see there it is. There's a text. CWC is working correctly. So we know that CoWors doing its thing. We're happy that even though we are going to create a file folder structure, we've already created a file and everything's working correctly. So in summary, in order to install CW, what you need to do is download the Cloud desktop app from claw.ai slash Download or you can go to claw.ai and just find the download button. And remember that CoWor does not work in a browser, so you have to download that once you've installed the app, what I want you to do then, you're going to need to click work in the left side bar, and that's a tab alongside the chat and the co tabs, but you want the cow tab. Remember that C Work downloads a virtual machine on first launch, which is what does some processing for us. I may not be a few minutes, but allow some time for that download. It's about two gig. And then remember that folder access is granted when you first run a task that needs it, and it's not during setup, so you need to actually run a task, and that will allow it to ask you at least for folder access. And remember that the desktop app must stay open while cow runs. If you're running any of the tasks and you close the desktop app, then your task will close down. It doesn't keep running in the background. So now it's your turn. It's time to get cow running on your computer. So what I want you to do is first of all, install Cloud desktop app from cloud.ai. Sign in with your pro or above account because as I mentioned, you need at least a pro account to do the work. Click the CW tab and then let the environment download, and this may take a little while. And then I want you to run the verification task that you saw me run to confirm that cow is working correctly. You don't need any specific files yet. We'll get to that at the right time. But once you've done this, you'll be confident that you've got cowork installed and ready to go. So have fun doing that, and I'll see you in the next lesson. 6. How To Grant Folder Access So Claude Can Read, Edit, And Create Files: In this lesson, we're going to talk about how to grant folder access so Claude can read, edit, and create files. Co Work can only act on files in folders you have explicitly selected. Nothing outside those folders is visible or touchable. This lesson shows you when and how folder access is granted and how to manage it. So let's get into it. So what we're covering today, the feature is folder access. Why we use it is to grant, cow, read or write access to specific local folders. What is solves is CW saying it can't find files or accidentally giving access to sensitive directories. So here we are inside of cow, and the first thing I'm going to do is paste in our prompt. And this says, create a setup confirmation file in my passion sports folder, create a file called rem dot TXT containing the text. Passion sports C workspace initialize. A list of subfolders we'll use orders, products, suppliers, reports, marketing, World Cup 2026, save it directly in passion sports. So this is actually going to create a file with this text in it. It's not actually creating the folders. So let's run that. It authenticates to begin, loads the tools. And then it asks Claude would like to work in a passion sports folder, so I'm going to allow that. It's noticed that the passion sports folder isn't found at root level, and it's asking me to confirm the path to the folder. It's give me a couple of examples. So I know that mine's within documents, so I'm going to copy that, paste that in, and then set it on its way. So now it officially asks if it can work in this folder with the full path, I'm going to allow it. The folder is now connected. It's creating the Readme dot text file. And it says that the Read M has been created at the location I've asked for with the message and all the subfolders, and it's ready to go whenever I want to start building out the workspace. So let's open up that folder structure now. And if we open the TXT file and zoom in a You can see it's written the message passion sports workspace initialized subfolders will be this. And although we haven't created these all yet, this is just a test to make sure that it knows what we are going to build and that it can write to the top level of the folder structure. So they've done that fine. So the key thing here is that we want to grant access to folders that contain the files that cow is going to work with, but we don't want to grant access to our entire home directory, the desktop, or the downloads unless specifically needed because the more access we grant, the more risk there are. We also don't want to grant access to folders that contain passwords, banking details, or sensitive personal data. And the key thing also to remember is that cow is going to ask for our approval before significant actions, but it doesn't have read access to everything in the granted folders. So be intentional. You can always add more folders when a task needs them, and you can do it step by step. So in summary, folder access is prompted when a task is needed. It's not preconfigured during setup. You can manage permissions via settings cow or via the operating system privacy settings. You can create your workspace folder structure before starting file based tasks and grant only what cow needs. You can expand access as the tasks require it. So now it's your turn. Why wouldn't you to do is set up your workspace with Inclaude. Create a dedicated workspace folder for your own work or business. Doesn't have to be the passion sports example. It's whatever is relevant for you. And you can start now adding subfolders for the type of work you'll do, reports, inputs, outputs, or whatever fits your workflow. You can run the read me task adapted to your folder and confirmed cowork come write to the folder successfully. So go ahead and do that and I'll see you in the next lesson. 7. How To Use Projects To Organise Your Work And Build Persistent Workspaces: In this lesson, we're going to go over how to use projects to organize your work and build persistent workspaces. Without projects, every co work session starts from fresh. Claude has no memory of what you worked on yesterday, what your preferences are, or what files it's already seen unless you stay in the same conversation. But projects fix this because they give Claude a permanent workspace for a specific job with memory instructions and scheduled tasks that persist across every session and every chat within that project. So let's get into it. So what we're covering today is projects, and why we use them is to group related tasks into a persistent workspace with memory, instructions, and scheduled tasks. And what it solves is starting from scratch, every session, re explaining context, losing continuity across recurring work. It's similar to working in a folder except that folder actually has context about what you've done. So here's a table with a picture of using projects versus standalone folder session. In other words, when you create a folder and you work inside of it on clawed cowork, that's standalone folder session. And then when we create a project explicitly in cow, that's what I mean by project on the right hand side of this table. So let's talk about it step by step. In terms of file access, the standalone folder session has access has file access, yes, and so does a project. So it's the same in that regard. Terms of memory across different sessions and different chats. For a standalone folder session, no, there's no memory. It starts afresh each time. Whereas with a project, yes, the memory persists automatically between sessions. With custom instructions, on a standalone folder session, no, you have no custom instructions for that folder session. So you can't say when you start working in a folder, always output and format in this particular way, and then it will remember it every time you work in that folder. Doesn't matter whether you've got that folder that you've worked in and you want to output a report the same way every single time or not, I will still forget what you did originally, and it won't remember any instructions that you gave it based on the folder alone. However, with a project, you can set it once, and that formatting will be applied every session because it's within a project. In terms of scheduled task, if you're working in a standalone folder session, the scheduled task will be global. In other words, you can see those scheduled tasks in every single folder session, and they're not hidden from you based on the folder that you're using. Whereas, when you start a separate project, you can create schedule tasks that are scope to that project, and you can only see them if you're within that project. And that's good for privacy reasons and to make sure different people can only see certain schedule tasks depending on the project that they're working. And in terms of context, if you're working on a standalone folder session, context doesn't get carried forward. So the context is relative to the chat that you're in, but not to that folder. It won't get carried forward to everything in that folder. Whereas when you start a project, then Claude builds on all previous work within that project, and it doesn't forget the context when you start new tracks within the project or do new work within that project. And all of this will become clear as we move forward. So let's talk about what's included in the project. So project includes the following instructions, which are standing rules for every session. They set the tone, the formatting rules, the background context, and then Claude applies them automatically on every task in the project without being told again. Then it includes also memory. So memory is persistent context across different sessions. Claude remembers what it worked on and what decisions were made. It's scope to the project only, and it doesn't leak into other projects, so it can remember things specific to that project and it won't get it mixed up with other projects, which is why it's good to have multiple projects for different pieces of work you're doing. And then projects also include context, and that's folders, files or URLs. You can attach a local folder and link a clawed chat project or paste a URL, and Claude references all of it on every task in the project because it's all seen as context, and also the things that you type within different chats within that project are also part of the context. And then scheduled task. So Schedule tasks are recurring pieces of work tied to the project. Schedule task inside a project will run with a full context and a memory, not as isolated one off jobs, which means the more you run schedule tasks inside a project, the more that they can remember what you did and used earlier context for their next run, which is really useful. Let's head over to Claude C Work and let's create our first project. So here we are in clawed cow. And the way to create a project is surprised to go up here to where it says projects. So first of all, click on projects. And then in here, you'll see that it says, Look, start a project, point Claude at a folder on your machine and work on it together. So we can get started by clicking New Project. And when we do that, we've got essentially three options. It says, create a new project, a dedicated place ongoing work where Context builds overtime. Files and instructions stay in a folder on your computer. So we can either start from scratch to set up a new folder with instructions and files, import a project, so bring a project you made in chat over to CW. So if you were using clawed chat up here, you may already have some contexts you want to include. You can use an existing folder. So we're going to go use an existing folder because we've got that all set up with some work done inside of there. So if we click here, then it says, use an existing folder, pick a folder and Claude treat its files and will treat its files as project context. Add instructions to shape how Claude approaches the work. So we're going to pick from the folder we created already, the passion sports folder. So we'll click here. Then we'll go to documents and then pick this passion sports folder and click Open. So now we can add some instructions. I'm going to add some instructions that I created earlier. And these instructions say we are Passion Sports, a fan first organic cotton T shirt brand launching into the World Cup 2026 market across UK, Brazil and Argentina. All outputs should be direct and business ready. Germany and Spain are out of scope. Never include them. And for now, we're not going to add any files because the files we're interested in are already in this folder. So everything's done, memory's on, we're just going to create a project now. So here we are. We're inside of our project. If we ever want to see where it is, we can go to projects over here, and we can see there's only one project at the moment. It's called Passion Sports. And we can double click in here and it will take us back into this chat. If we ever want to edit any of the details here, we can go up here, click on these three dots and go to Edit details. And I'm actually going to rename this because the work we're doing is a bit more specific than just passion sports. So I'm going to call this project Passion Sports for World Cup and then click Save. So there it is. And again, that will exist inside of this project's menu item on the left. So now we're going to past in a prompt just to test the way the project is working. And this prompt says, read everything in this folder, summarize what you know about this workspace, what's here, what I use it for, and what instructions you'll follow. If anything is unclear, ask before proceeding. That's completed after a few seconds. It says it's read everything. And what it says is what passion sports is, it tells me my name. I'm the founder of passion Sports, independent British FanF Murch brand making 100% organic cotton football t shirts. And the brand sits between cheap license merch and expensive replicitst argeting football fans age 25 to 45. So this is all from reading what's in my folder and things that I've generated based on previous work sitting inside of the passion sports folder. And the World Cup 2020 launches where things stand says the tournament runs 11th of June to 19th of July, which is less than a month away. And my core range targets England, Brazil and Argentina fans. It talks about the current approved hero products all noted here. And then it tells us about the suppliers that we're working with, the marketing, the budget, the channels that we're promoting on, and basically all of the stuff that it can find inside of the folder with key things to watch out for and what rules it's going to follow while we're working here. So that's just to test and give us a complete overview of what's going on here. Now, to show what shared context means, all we need to do is create a new chat within this project. Or actually a new task within this project, and it will always reference the information and the context within this project. So in order to do that, we can go here and click New task, and we can make sure that we're working within a specific project, which is Passion Sports World Cup. And now that's ticked, we will be using the context from this project. So now we can ask it questions like prices about our T shirts. So I've typed in. What are the prices of each of our T shirts? Let's run that. And after a few seconds, very quickly, it says, based on the pricing file, here are the recommended retail prices for the in scope T shirts, UK, Brazil, and Argentina markets. So here you go. Here are the prices, and it's picked all of that up because we've created a project, and we've based that project on the files in the file system. So if I open up the files in the folder, and all we need to do is look within passion sports and then products World Cup 2026, and you'll see that there are some spec sheets here and some other pieces of information to do with suppliers. We've got invoices here, we've got reports, and all of these things are being used currently to pull the information that we're seeing within CW that is leading to these prices. So the idea is that as we work through this lesson, we will build more and more files, and then we will see that this information will be able to update within the project, and anything within our company that we work on that exists within this project will be at our disposal if we want to ask any questions, and that is shared context. That's what makes life easy for us because we don't need to repeat ourselves. It automatically knew what we meant when we spoke about T shirts. So let's talk quickly about how memory works in projects. Memory is enabled automatically for cow projects. Claude remembers context from task you've run in a project and applies it to future tasks in the same project. Memory is scoped to the project. What Claude learns in your passion sports project or any other project you create stays there. It doesn't carry over to other projects or standalone sessions. And this is why the same task brief produces better results inside a project over time. Claude isn't starting afresh, it's building on what it already knows about your work. And the current limitations at this time is cow projects are desktop only and stored locally. Projects live on your computer and there's no Cloud sync, so you can't access a project from a different machine. It's cow only. Projects are not available in clawed code. There is a version for clawed chat, but it's a slightly different thing. So in terms of agentic or AI agents, clawed Cowork has projects, and it's not supported yet in clawed code. However, support is planned for a future update. So by the time you read this, it may well have been available in clawed code. Archiving doesn't delete your file. So when you archive a project in cowork, the metadata is removed from the user interface, but your local files and folders are untouched, so you still got all of that on your file system. And then in order to use projects, you'll need the latest clawed desktop because Projects is a brand new feature at this time, so update to the latest version of clawed desktop app before trying to access projects if you haven't recently. So in summary, projects give Claude a persistent workspace, memory, instructions, contexts, and scheduled tasks that carry across every session within that project. There are three creation methods to create your project to start from scratch, import from a clawed chat project, or use an existing folder. Memory is scoped to the project, and Claude learns your work without leaking into other projects. And instructions set once are applied automatically on every task in the project. There's no need to re explain yourself, which is part of the power of projects. Also, these projects that we're talking about are available on the desktop only. They're stored locally, so there's no cloud sync at this time. It's your turn. What I want you to do is create your first cow project. Go to Projects and then the Plus button, use an existing folder and then select your passion sports folder or your own work folder, if you prefer to create a project for it. Add a short instructions block, which should be your business context, the tone preference for the way cowork will speak to you and type to you. And anything Claude should always know about your work for that project. And then run your first task. You can ask it to read everything in the folder, summarize what you know, and ask if anything is unclear, and then Claude Cork will confirm to you that it understands your context. And from now on, run all your tasks inside this project, and it will save you a lot of time. So have fun doing that and I will see you in the other lessons. 8. How To Set Permissions And Stay In Control Of What Claude Can Touch: In this lesson, we're going to talk about how to set permissions and stay in control of what Claude can touch. You see, Cowork is designed to keep you in control before it takes any significant action, renaming files, creating folders, modifying documents. Before it does any of that, it shows you a plan and waits for your approval. And this lesson shows you how to use the approval flow and how to steer cow when it needs adjusting. So let's get into it. What are we covering today? Today, we are covering task approval and oversight. Why we use it to review and approve Cowork plan before it executes. And what it solves, it solves Cowork making changes you didn't intend or feeling like you've lost visibility. So the scenario is it's week one at Passion Sports, and the founding team has a rough notes file with goals, product ideas, target markets, brand values, and you want to turn it into something visual that the whole team can see and react to. And that's one brief one HTML page that opens up in any browser. So this is the first time that the T shirt concepts appear in this course, and that will be England, Brazil and Argentina T shirts shown in their proposed different color ways with design direction notes. So these are the notes that we want to transform. What they look like at the moment is there's some notes from the founder of the company. That would be me. And it's the mission saying that we're going to make football march that looks as good off the pitch as on the pitch. The high street is full of cheap licensed products and overpriced premium kits with a brand that sits between them, well made, actually designed, price for fans. Here. And we're making this for football fans age 25 to 45. So we've got in this brief exactly who we're making it for, and there's some more information there. And then the brand values, quality first, fan perspective, honest design, small but proud. Then what are the product pillars, material quality, design integrity, fit and sizing? And then the World Cup 2026 is the opportunity. It talks a little bit about the tournament, who's hosting some of the key dates. And then it speaks of what our hero T shirts are for the tournament. So these are the main ones we're focusing on the English T shirt, Brazil, Argentina. Where we are now, we've got a number of email subscribers, Instagram followers, and we've got a current we've got some current stocklist conversations with some stockists and some supplier relationships with these companies here. Here, down here, we've got what we want from the launch, so to sell 1.5 k units, to get at least one piece of real press coverage to grow the email list to 15,000 and to prove the model so we can essentially scale and plan a bigger campaign. It's not about getting rich from the tournament. It's about building a brand that's going to last five years plus. So these are the founders notes, and what we want to do is we want to turn this into another form, so let's head back to Claude, and then we'll see what we're going to do with this text document. So here we are back at Cowork and we've paced in our prompt for what we want to do with our founders notes. And what it says here is read my Slash Passion Sports Passion Sports notes, TXT file and produce a visual HTML brand vision board. Is going to be a vision board that visualizes the following information. The board should include a mission statement card, which is extracted from the notes, a target customer card, three product pillars, which is what makes our t shirts different. World Cup opportunity summary, first visual concept of our hero t shirts, a short design director note, which shows each in its proposed color with a short design director note, England Brazil and the Argentina and it's saying exactly where to save it. As usual, at the end, it says, show me your plan before creating the file. So let's get that done. Here we go. We've got the plan from Claude, and the plan is for the Passion sports brand Vision Board. The layout is a dark themed full width HML page with a bold header, then a grid of cards, clean premium fill, streetwear adjacent, not corporate. And the sections will be the mission statement card, the target customer card, three product pillar cards, World Cup opportunity summary, and the hero T shirt visual concepts. And the visual style, as I said before, is going to be a dark background, white or gray text, colored accent borders. The mock ups will be styled in SVG shapes with the actual color showed. And then it asks, should it go ahead and build it? So that looks pretty good, but I want to add one more line, so I'm going to paste in one more prompt, and it says, That looks right. Add one more card, which is our key brand values extracted from the notes, then proceed. So let's get that started. So now, supposing while it's working, you discover that something's wrong here. You can actually at any point, stop in the middle of the flow, like so. And let's say we discover looking through it, we just take a look at what we've said for the different t shirts. We've said Argentina T shirt is sky blue and white, but actually, it should be sky blue and white stripes. So let's correct that allows us just by stopping it, to be able to go back and correct anything even while it's running. So the T's colors look wrong. Argentina should be sky blue and white stripes, not plain blue. Fix that before continuing. And then we keep going. So now our vision board has been created. There are actually a few ways to see it. We can click here where it says View Vision Board, and we see it opens up on the right. We can also click down here that would have opened it the same way. And just to double check, let's go in our file system, and we'll see exactly where that's been generated because that's the main time saving thing with Claude Cowork is that it uploads files itself, it picks up files from your file system, and it also downloads them for you without you having to do anything and a lot more. So let's have a look. Here we are in the file system, and you can see we're at Passion sports and here it's created the Vision Board HTML. So this is something that's created as a web page that we can actually open up in our browser, so let's double click on that. We can see here's our vision board, and it's the Passion Sports Vision Board. It's got the mission statement. We make football march that looks as good off the pitch as on it. The high street is full of cheap licensed products and overpriced premium kits with a brand that sits between made, actually designed, price for fans who care. So there's the mission statement. We've got information from our text file, which didn't look as colorful before, now it's got some color to it in terms of, we've got the fan, who cares? And this is our prime customer, primary customer, and we've got some tags down here about them. If we zoom in a little bit, we can see some of the tags that we've got about the fans, and then we've got our brand values here, which we added, which are the last thing we added, quality first, fan perspective, honest design, small but proud. And material quality, design, integrity, fit and sizing. So talk about how they qualified when the tournament kicked off, facts about the company that we saw in our text file. And then we've got the very first visual concepts of the T shirt. So it's just the beginning, obviously, improvements to be made here. But you can see that just by specifying where the file is, it's generated a new file for us that could go on a webpage. And if we want to make any tweaks to it in terms of color, for example, this looks quite dark. We can always go back to Claude. And update it. So let's do that. So I've pasted in this prompt to update the vision board slightly saying, give the vision board a white background and modify the text color appropriately. In sure all visual concepts are visible on white background, give them individual backgrounds if necessary. I've said that because obviously, we've got some Ts that have a white background themselves. Maybe it'll keep these darker or maybe it'll change them. Let's see what it does. So here you go. Here's our newly updated vision board. I think it's done really well. It's changed background colors where necessary. And if we look here, you can see it's changed on the file system as well, so we can open that directly as an TIMO webpage. Looks a lot brighter. I think I like this one a little bit more, but I think I might change the background. The beauty of it is that we've done all that within clawed Cowork. We didn't need to touch any files whatsoever. It's all generated for us, and we can visualize it in here and see it on our desktop, upload it to Google Drive, if you want, whatever we want to do, it's all done. So in summary, I recommend that you always include show me the plan first in new or complex task briefs. Remember that Cowork will wait for approval before significant actions, and that's by design, especially because it's on your desktop. Remember that you can redirect, refine or add to the plan before approving, and typing stop or pressing the stop button at any point immediately halts execution. You can do that whenever you want to change something. So now it's your turn. What I want you to do is practice this approval flow. I want you to run a task on a non critical folder for your own work, read the plan line by line before approving, and then change one thing about it first, and then get comfortable steering before using cow on important files. And by steering, what I mean is updating your plan, changing your plan, reviewing it line by line before you do anything more. And this will help you when you're actually doing much more important work. So, have fun doing that, and I will see you in the next lesson. 9. How To Assign A Task And Let Cowork Handle The Steps: This lesson, we're going to talk about how to assign a task and let Cowork handle the steps. This is the core cow skill, giving Claude a multi step goal, and letting it plan and execute each step without you managing them one by one. So we're running Passion sports first real planning task here. So let's get into it. So what we're covering today is multi step task execution. Why we use it? Because Cowork plans and runs a sequence of steps to reach a defined outcome. What it solves is manually doing multi step work step by step when Cowork can handle the whole sequence. So the scenario is that the Passion sports team has agreed the World Cup 2026 strategy. Now you need a visual project plan that the whole team can follow because this is the way that teams can essentially gather around a plan and then work to that plan, keep everything on time. And that's going from design through to the actual tournament week, so all the way to actually carrying out the plan for the World Cup. And so you have a brief text file outlining the phases and key dates at the moment. But the output that we want is a color coded HTML gant chart for you project managers out there, you'll be familiar. That opens in any browser, and it's got five phases, four milestone markers, and territory launch dates flagged. And it's also got one brief visual output. So let's get into it. So here we are back clawed in the Cowork tab, and we're going to paste in our prompt. And before we go through what we've pasted in, we're going to notice that we're going to be reading from the exact same text file, let's have a quick look at that text file. So as well as this file giving us an overview of the launch brief, having our product range, it's also got a number of launch phases. And what we can do with this very same file with key milestones and launch phases is build a product plan, and we're going to now use Claude to take the exact same file and build that product plan and put it in the right place on our desktop. Our prompt says, Read my passion Sports World Cup launch Brief, textFile and produce color coded HDMPduce a color coded HTML ganchart for the World Cup 2026 launch. The chart should cover January to July 2026 and include phase one, design and product sign off, phase two production supplier orders. And it essentially goes through all of the different phases of what we need to deliver right through to the tournament window and live sales because we expect to be launching and selling during this time period. And it also says, to mark these milestone dates where all designs are approved, all stock is with KI Pro, which is one of our suppliers. The marketing campaign is live, and then the World Cup kicks off. Use a different color for each phase, highlight England, Brazil and Argentina launch dates, and then save them all back to this World Cup Gant chart, World Cup HTML file. Show me your plan first. So let's do that. So here, Cowork confirms that the file it's going to generate is this World Cup Gant chart HTML file. It says it will create a report subfolder if it doesn't exist. And then the layout is going to be a white background HTML page, probably because that's what I asked it to do before a fixed head row of the months and then rows of color phased bars and a separate milestone year below. So here it's telling us what phases and colors it's going to create for each different phase from design and sign off all the way to the tournament win. Then it's got milestone markers in diamond on the timeline, so you'll put little diamonds in whenever it reaches that correct place on the timeline for this particular milestone. For example, when all designs are approved or when the players purchase orders are raised or when the tournament ends, and especially when the World Cup kicks off, special marker for that. It's also going to add the country launch dates, as we've asked it to, and it's got highlighted flag markers on separate subros. And for each work stream, the sub rows within each phase are pulled from the brief, which is great. Additional detail, a small legend on the bottom and a budget snapshot pulled from the brief. So it's going to add that so that we can actually tell what colors mean and what the different sections of the plan look like. Shall I go ahead and build it? I'm going to say yes, and then we can crack on and see it do its thing. So let's go. So here we go. Here's our finished gunshot, and just to check just to make sure we can check here and see that within the reports, we've actually got the Gan chart in the file system so we can double click and open that in HTML format, which means that the whole team can see this and we can update it as we need to. See if we zoom in slightly, we've got the title here for the World Cup, 2026 launch, our budget, the number of SKUs or products, the tournament start and end date, the target, number of units sold. And then we've got the key for our phases here so we can see that these colors relate to these colors. So this is phase one, design and sign off. Production and suppliers here, marketing and build phase four, which is pre launch and sell in here, and then tournament and live sales here. And then key milestones are marked in yellow here. So you can see, for example, the World Cup kicks off here 11th of June, and that's marked in yellow. But there are different colors for various milestones. So you can see here the designs being approved by this date, suppliers purchase orders raised by this date. Stock should be at one of our suppliers by this date. So if I zoom in slightly, can see that stock should at our suppliers there, we have various emails that we need to send on different dates and this says, by 15 April, the first email should go up there. Pre orders are open 1 May, and the campaigns live on the 15th before kickoff on 11th of June. We got a little bit muddled here, so that's something that we could go back and correct within Claude. It seems to be an overlapping of the different flags here because the tournament window open, so we can go back and reposition that. And then the tournament ends on 19 July here. And then if we look down, we've got all our budgets here nicely stacked on top of each other with the total. We've got the different workstream owners. So these are just some generated names for the different work streams that we see here. And then we've got a risk log of any possible problems that we see that may be coming up, and we call those risks. So stock delays, for example, or low self through or early tournament exit, what we would need to do, such as create a generic range, which covers the downside. So this is quite a comprehensive plan all generated by Claude and all put on our file system directly, so we don't have to think about it anymore than. Part of the beauty of this is that we can upload it, we can update it without having to go into any tool. So let's say we're going through this and we actually realize that one of these phases is going to end on a different date because obviously plans change. If we want to update this, we can actually go back to Claude Cowork and we can do that. So we go back into Claude and I'm going to say phase three bar looks too short. Marketing should run through to the end of May, not just April. Please update and resave. So we're going to now run that. And if we look over on the right here, we can see that actually Claude updated it so that this bar now goes through to the end of May. And it was as simple as that. This actually took just a few seconds. You can see it's doing some calculations here to find out how long this bar should be. So within a few seconds, we've got now an updated plan. Excellent. So in summary, when you're writing your tasks, write them as a numbered list with a clear output file defined, and you want to include the folder path and the output file name explicitly, then you want to look through the plan and approve the plan, let cow run without managing each step, and then finally review and refine in one follow up message if needed. So now it's your turn. What I want you to do is run your first real task and then pick a real multi step task from your own work, something that normally takes 20 to 30 minutes. I want you to write it as a cow brief with numbered steps and a defined output file name and then approve the plan, let cow run and review the output, and then refine in one message if needed. So have fun doing that, and I will see you in the next lesson. 10. How To Organise Messy Folders Using Cowork: In this lesson, we're going to go on about how to organize messy folders using Cowork. Folder chaos slows every business down. It's files names, strange names like final underscore V three, underscore use T dot CSV, spec sheets mixed with random screenshots, design notes, sitting all in downloads and various different downloads, maybe for software, all different types of things. What Cowork does and how it can help you is to read your folder structure, propose a new structure, and reorganize everything after you review and approve. Then it uses the clean files to produce a finished output if you want it to. So let's get into it. So what we're covering today is folder organization. Why we use it is so that Cowork can read, propose, rename and restructure your files. And what it solves is spending hours manually sorting files that Cowork can handle in minutes. So here we are in Cowork, but before we do anything else, let's have a look at our messy folder and how we're going to clean it. Here's our folder. This is the Downloads folder, and it's with impassion sports. And so if you have a look here, you can see we've go various different types of files. They're all named differently. There's no particular pattern to how they're named. So, for example, we've got Argentina approved and this file if we open it up. This file has product details, so it's got the Argentina fan t shirt at the skew, which is a number that uniquely identifies this product, the color, sizes, material, and things like that, recommended retail price. So information generally about the product. So that's what this file represents. If we go back to our folder, can see we've got a few files like this. So not only have we got Argentina, we've got the Brazil Training T shirt, we can look through and see we've got English T shirt, specs in here as well, all named completely differently. And then we've got things completely different like we've got this temp file here. It says, TEM, delete this. So I don't know what that is, but we definitely don't want to keep files like that here, and we want to make sure everything's named correctly. Another one here is we've got the World Cup campaign brief, so if I open that one, you can see here this is the walk up 2026 campaign brief version three final, and it's got the brand, the campaign, the name of the campaign here, which is the fan T shirt range for Work upp 2026, the period that's running in, the objective, and why we're doing this, the key message, and what channels we're promoting it on budget and the lead territory, which is UK. So you can see that all these files are completely different and they all live within this downloads folder. They're all named completely differently. Some of them don't even need to be here like this random stuff TxD. So we've got stuff we need to clean up, stuff we need to rename and reorder. So let's go back into Cowork and see how we can achieve that and save ourselves time and hassle. So here we are in Cowork, and what we want to do is we want to paste in our prompt that's going to help us to clean up our messy folder. So here's my prompt. It says, help me organize my passion sports downloads folder, scan the contents and propose a plan. What categories or folders should I create? How should files be sorted into them? What naming convention should I apply? Any files to flag for review or deletion? Show me the plan and wait for my approval before making any changes. So let's run that. So that started up, and Cowork says it's working on it. So Coors come back now, and it seems to have done what we've asked for. It says, There are 25 visible files plus the hidden dot ds store file. Here's the full plan. All 26 items accounted for nothing touched until you prove great. So it's given us the folder structure. It's got a design folder for product specs, design notes, et cetera. A suppliers folder for pricing and rates from all three of the suppliers that we picked, a marketing folder for campaign briefs, emails, et cetera, and social plans, a notes folder for meeting notes, action items, and loose ideas, future products folder for specs for the teams outside of the core World Cup 2026 range, because remember, we've got a core range and then some other products that we may not release. And then an archive folder for superseded draft drafts that we took notes with or created a document with, but then they've been superseded because we created a new version of it. They'll go in here because we're not really using them. That seems fair enough to me. And then all 26 files because that's including the hidden file, where each one goes. So in the design folder, it's giving us an example of how it's going to do the renaming. So, for example, instead of England space Ts, Base specs, space final tXT, it's got this, which is a lot tidier. So it's got England hyphen fan hyphen T, hyphen spec DTXT, and it keeps to some extent that naming conventions. You can see for the Argentina design, it's named it Argentina fan t approved DTX because it happens to have approved in it. But it's these hyphens. The fact that it's a fan t shirt, it includes in there, or if it's a training T shirt, it includes that in there. And the only time it includes versions is if we've put it in there, so it'll say V three. But it's a lot cleaner. It's all lowercase text, so I like that. So essentially, it's going to rename things so that they're a little bit more consistent. And then for the suppliers files, it's going to do the same thing. It's put in hyphens. Starts with the name of the supplier, a hyphen whether it's pricing or not, pricing related or not. And then the date if it's got the full date, if not, you'll just put the year in, but at least you can see here it looks a lot easier, more easily readable. And it will do essentially the same thing for the marketing documents and for the notes and for the future products and for the archive. And then there are three files to be deleted, which is the DS store, which is a system file of no value, the temp file, which actually says, delete this and the Germany spec draft, which says do not use anyway. So those can be deleted. It says the naming convention is Kebabcase which I love the sound of, and it's lowercase hyphens, no spaces, not all caps, no parentheses, and no status markers like final approve use this because you probably know from using files, these always change. Final is not necessarily always final. So it's removed those folder location carries the meaning, so the folder will tell you whether it's approved if you want to create an approved folder. So all 26 of those files have been placed. It says, Say the word and I'll execute. So I pretty much like that plan. I have nothing that I want to improve except I want to send these files to a different folder. So here's the prompt I'm going to put in. So I've said, great, do that. Then once you've tidied up the files and put them in subfolders under downloads, move all the subfolders under slash Passion Sports Products World Cup 2026. Apart from that, I'm happy with it, so let's go ahead and run that. So Cowork finished doing that. Now, it says 23 files sorted, now moving the six folders, passion, two products, World Cup 2026. The errors just mean that the World Cup 2026 already had those folder names present from a previous session. So I already created that folder, so no worries there. But all 23 files landed correctly. The structure is exactly right. All done, two things to note. Three files couldn't be deleted because of sandbox permission, so I didn't really give it permission to do deleting on my file system, so that's okay. So inside of downloads, we're still going to have this temp file, the draft that says, Do not use and the DS store file. I'm not sure we should be deleting that anyway, but it was a test. It says, Please trash those manually in the Finder and the downloads folder will be clean. Everything else is in Passion Sports products World Cup 2026, and it's split now into six subfolders. So let's go on and have a look at our folder structure and make sure it's done what it says it would do. So here we are in our Passion sports folder. First of all, let's have a look at the downloads folder. So in our Downloads folder, we can see that we've only got now these two files, which it couldn't delete, and I'm going to delete them myself. So there you go. They're gone. And now you can see here within this World Cup 2026. Win this World Cup 2026 folder, you can see it's our new folder structure. So we've got an archive folder with the old pricing, the screenshot, and then the old brief in there. We've got a design folder with all of these product specs for the different T shirts and image reference, and then some design notes. They all relate to design. We've got future products for Italy, Korea, and Spain in here. And then we've got marketing documents in here, campaign brief, email, and social media plans to do in marketing. We've got all our notes in here, follow up actions, miscellaneous ideas, random notes, and retailer call notes. And we've got information about all our suppliers in here with the pricing across all suppliers, all with the new naming conventions. So that's a lot neater, a lot tidier. Much happier what we're going to do now is we're going to produce a lookbook, which is a look at all of the different types of T shirts in SVG graphics. So SVG graphics are just really simple plain graphics because that's what Claude does, and it's going to show us all of the T shirts in our range with some information. So I've pasted in this prom which says, Now read all the files Impassion Sports Products World Cup 2026 and produce an HTML product range lookbook. And I want to show each T shirt as an SVG flat lay in its correct colorway, so that will be an image with the correct colors. Include the skew, the SKU, which is a unique identifier for each product, the recommended retail price, the territory and the launch status. Group it by territories, UK, Brazil and Argentina. So any items related to each country will be grouped in there, so you'll see t shirts related to each country. And then save it all in passion sports market Passion Sports book dot HTML. And this gives us an HTML file we can either put on a website or just click on and look at anytime it gets updated. It makes it really easy. So let's run that. So here we go. Claude Cowork says that it has created the lookbook. So let's first check in our file structure to make sure it's put it in the right place. So we said it should be in Passion Sports Marketing Passion Sports lookbook dot HTML. So let's have a look, and here it is right on time. So let's open that up in Chrome by double clicking. And here we go. So this is World Cup 2026 Lookbook. There are a few things that could have been brighter in color, but that's something that we can always improve essentially at the top, it's giving us a little bit of information such as the fabric we're using, the key location and the price range. And down here, you can see our T shirts or the first look, and these are really basic ones. We're not saying we would use these necessarily in real life, but this is to show you what can be done. So it's created our lookbook for the United Kingdom. We've got the England 2026 World Cup tea, and if I zoom in a little bit, you can see we've got the price, the territory, which is UK, approved for production, yes, and the sizes, the size range is four color screenprint, and we're going to order 500 units. Got the same for every single one of these t shirts, the Morocco weighty, and so we've got the same thing there, same information. And the same for the Brazil t shirt, and this is the training T shirt. We've got the Argentina fan t shirt, and we've got the Spain classic T shirt. And in development, so these have not been approved yet, are the South Korea fan T shirt, the Italy classic T shirt, and the German training So here we go. This is our lookbook and this will evolve as we change things. No doubt. We've all changed them. And we can just say that the good thing is we've created this all without even touching a file system just from a few prompts on Claude and just from some files where we tied up our messy folders. So that's great. So in summary, if you want to organize your messy folders, you need to describe the folder and ask for a proposal or plan and review that plan before approving. Then Cowork handles the renaming, the restructuring, and the creating of the subfolders. And then any files it can't categorize are flagged for your review. It doesn't delete without your permission. And then you can chain tasks together, so you can organize the folder, then immediately use the clean files to produce the next output simple as that. Now it's your turn. What I want you to do is tame or tidy up your messiest folder. I want you to pick the messiest folder in your own work or your business, ask Cowork to scan it and propose an organization plan, and then review it before approving. And then I want you to run this on the files that you can afford to reorganize, then chain it to produce an output from the cleanest files. 11. How To Turn Raw Data And Screenshots Into Finished Spreadsheets: Lesson, we're going to talk about how to turn raw data and screenshots into finish spreadsheets. A lot of business data arrives in messy forms, order exports with columns you don't need, screenshots of pricing tables or maybe just pricing tables inside of textiles, supplier spreadsheets in different formats, a whole load of different files in different formats. And Cowork reads all of these and can produce a clean, consistent spreadsheet that you can actually use. So let's get into it. What we're covering today is data extraction and spreadsheet creation. Why we use it to read multiple source files and produce a clean output spreadsheet. And what it solves the problem where you're manually copying data between files, reformatting CSV files and merging spreadsheets by hand. You don't need to do that anymore. Claude can do it for you without even touching a file. So before we do anything, let's go and have a look at our file system, and we'll see the kind of files that we want to read and clean up. So here we are in our Finder window in our file system. You can see within passion sports, we've got this orders folder. And if we look in Week 18, there are a bunch of different order files here. So let's open one of these up. And we'll see what it's all about. So in this order file, you can see there's orders for various products. We've got an order ID, an order date, customer ID, customer name, customer email, phone number, got address details here, both billing and shipping address details, the kind of stuff you'd expect to find in order form in order spreadsheet. And you've got unique identifiers, product category, price, all the normal stuff you'd expect to find. And obviously, we've got a lot of different columns in here. Sometimes we want less columns or we want to do something different with the data. There's also no totals at the bottom, so we might want to sort that as well. We don't necessarily want to have to go into every file and do that, though. And if we come back out and we look at our file system, you can see there's at least three files in here, and going forward, there's going to be a lot more files all with different data, but we might want to apply the same format. Also, within our suppliers folder, you'll see we've got this description from the supplier. So if I open that up and zoom in a little bit, you can see this is from one of our suppliers, and it tells us the pricing of some of the different products that we're buying from them. And in here, you can see there's a pricing table, and it's priced for different products, the minimum order quantity, the price per unit, and the lead time, how long it will take before it's delivered. And you can see it's for different colors of the cotton t shirt product and then the premium blank product in white or color as well. It's got the different prices there. So these are things that as a company, we might want to buy. It tells us the discount if we order more than 500 units, payment terms, all that kind of stuff, and then the contact a piece of contact information. So this is information that we want to get maybe into a different form, but currently, it's in this text file. Also, if we go back to the suppliers folder from one of our other suppliers, we've got an invoice. So if we open that up, you can see that it's a standard invoice with the invoice number, bill two, bill from and pricing details. And down here, it's got the payment details and the status. But the kind of stuff you'd usually see in an invoice. So no surprises there, but the main thing is it's in a different format. This one is in PDF format. And sometimes we want to take multiple different files, different formats, and then put them all in the same format or do something to all of them at the same time. What we don't want to do is we don't want it to be tedious. We don't want to have to go through our file system in order to do that. Let's head back to Claude and see what we can do to tidy up all these documents using the power of AI. So here we are back in Claude Cowork. And first of all, what we're going to do is we're going to tidy up the files that we saw in our orders folder, and we want to combine those into the same format. So here's our prompt. So what I've said here is read the three CSV files in Passion Sports orders week 18, and these are the three files and create a single spreadsheet in passion sports reports weekly sales, so that we've got everything in one place in one spreadsheet and then not CSV anymore. And include only these columns, the order ID, the date, product name, the quantity, price, status, and customer email. So as you saw, there are quite a lot of columns in that spreadsheet or in the CSV files that we saw, and we want to make it a lot more concise. Add a summary wrote at the bottom, so we want to see the total orders, total revenue, and average order value, and then highlight any orders with status equals canceled in red, so we can see any orders that got canceled. And before you do anything, show us the plan first. So let's run that. So here's the plan, and coworkers confirmed it's found 15 orders across the three files. And what it's going to do in order to create this Excel file is it's going to take this source data. It's going to extract the data for just these columns, the ones that I've requested. It's told us how it's going to format the date and the final price, including VAT for any discounts. It's told us how it's going to format with the head arrow in bold, light gray background, the price in pounds, and the date column formatted as day month year, but in this format. And the column whips are also fitted to the content, which is good. It's also saying that the canceled orders are going to be highlighted in red, and it's pointed out what they are. And the summary row at the bottom is bold, top border to separate it. And there's 15 orders in here. It wants to confirm one thing, which is shared revenue and average exclude canceled or return orders. It's defaulted to including all of them, but I'm going to strip those out. So I'm going to make a couple of updates here. So I've added a prompt to say that the revenue and the average should be excluded and to continue with the plan. So let's go. So Claude's completed. It says it's done all 15 orders across the three CSVs exactly as it said, and it's stuck to the plan here. So let's have a look on our file system and see what it's done. So in our finder window, if I hit Reports and then look at this weekly sales spreadsheet, we can open that up, and here it is, so let's zoom in a little bit. And we can see just as we've asked for, we've got the order ID, date, product, quantity, order value status, and customer email. And we've got three orders that have been marked in red. So these are the orders that have been either canceled or returned. And then we've got a total, and that total should not consist of the three orders marked in red. And so what I've done is I've highlighted just these cells, and I've not included these three. And if I look down the bottom here, I can see that the sum is 691. So that's exactly what we expected 691, 43, so we can see that's totaling up correctly. So as you can see, we've got a much tidier spreadsheet, and we didn't even have to touch the file system. We just told Claude Cowork where to look. What to do, where to produce it. And now we can just go back to our Find down. We can see there's a new file created. Excellent. The next thing we're going to do is we're going to deal with this supplier information. So we've got this file here that I showed you the pricing description file, which is just a text file. It happens to have this table, but we want it in a better format. So let's go back to Cowork and sort that out. So I'm going to paste in a prompt here, and it says, There is a file called Fabric First Pricing description in passion sports suppliers. Extract the pricing table from this file and save it as a properly formatted spreadsheet at this location. So I'm just going to add on the extension here. So I've corrected that, so it's got the proper extension, and let's run it. So we can see that it's now said that all seven product lines across the two ranges, organic cotton blanks and premium 2020 GSM, plus the volume discount payment terms and contract details in a note section at the bottom. So it's just saying that it's generated the file, and it's done some extra stuff at the bottom for us. But let's go look at the file that's generated. So if we go back to the suppliers folder, we can now see that there's this new spreadsheet that's been generated. Let's open that. And if we zoom in a little bit, we can see that it's taken the data from this text file, and it's made it into a nice spreadsheet for us where we can apply formulas and do whatever it is we want to do. We can see that the information is correct on both sides. It's got the same number of rows, and it's got also the volume discount, payment terms and contact that we've got on the left here as well. Giving it a nice title, it looks much clearer now. And this is actually in a format we can do a lot more with and looks a lot more professional. The next thing we want to do is we want to look at this invoice PDF that we saw earlier, and we actually want to take information out of here, put it into a different form, and actually just extract specific information that we're interested in from this. But we want to do all that without actually touching the PDF because we would need to copy and paste things out of there. So let's go into Claude Cowork and do that. So here we are again, and I'm going to paste in our prompt. So here we've said, I've uploaded a PDF invoice from our supplier print force. Extract the following from it. Invoice number, line items with quantities and prices. Total amount and payment due date and then save the extracted data in this file inside of the supplier's folder. So we're just going to extract the information that we're interested in and put that into a spreadsheet. Let's go. So I put in some wrong word in there. I said that I had uploaded a file, so I've corrected my prompt to say I've put a PDF invoice from our supplier print force in this location. And as there's only one PDF, it should have no problem finding it. So I've corrected that and let's run it again. So there you go. Cowork got everything from the invoice and built the spreadsheet. So let's go to our Finder and have a look. So we can see we've got this file called Print force Invoice Extract April 2026. Let's open that up. And if we look through here, we can see that it's copied the relevant information. It's got invoice number, date, due date, payment terms. And in here, it's copied just what we need here, the screen printing, Brazil training T. So this product here and the setup fee, and it's copied then perfectly 185 and 45. Also got our subtotals, the VAT, value added tax, and the total due. So we've got all the information we wanted, but again, this is now in a much better format, and we didn't need to do any copying and pasting from this PDF in order to do it. This is saving us a lot more time, helping us to be a lot more productive, and obviously, that means we save a lot more money. So in summary, Claude Cowork can read your CSVs, Excel files, images of tables, PDFs all in a single task. You specify exactly which columns you want, and Cowork can strip out the rest and take just the information you want. Things like summary rows, conditional formatting, red highlights all supported by Cowork to make life easier and if you want to, you can use the plus button in Cowork to upload files directly into the task, although you usually don't need to because you can tell it where to get the information in your file system, in your folders, and it does all the work for you. So now it's your turn. What I want you to do is clean up some of the data you've got and change it to different formats. So take a CSV or a spreadsheet or anything, a PDF, text file from your own work that has more columns than you need. Ask Cowork to extract only the columns or the information that you care about and then produce a clean output file, and then add a summary row and highlight any rows that meet any condition you define, and it'll be a lot clearer and a lot cleaner for you all with the power of AI. So have fun doing that, and I'll see you in the next lesson. 12. How To Generate Reports And Documents From Your Files: In this lesson, we're going to go over how to generate reports and documents from your files. One of the most powerful Corp capabilities is turning scattered files into finished readable documents and dashboards, and we do like dashboards. Data files become briefings, invoice files become color coded status boards, much more visually engaging and much more helpful to our teams. This lesson shows you how to go from raw inputs into polished output in the form of dashboards without any manual writing. So let's get into it. What we're covering today is document generation from files. Why we use it is to read source files and produce structured finished documents and dashboards. What it solves is spending an hour writing a report that Corp can draft in minutes from your own files, and also it saves time creating HDML dashboards from scratch that you can just use AI to generate. So the scenario here is we're in mid April at passion sports. Production is underway across three suppliers, Kit Pro, print force, and fabric first. Those are our suppliers. And we need a live view of where every supplier stands at the moment, all of the outstanding invoices, upcoming due dates for the product lead time status, so we want to know how long things are going to take in general, including shipping. And we want to do this all without opening a single spreadsheet. So the output we want is an HTML supplier status dashboard. And in each supplier, we want to see as a color coded card that opens up in any browser, and then the operations team can see everything at a glance in second so we can see all of our suppliers what they're producing and how long it's going to take for them to produce and get all of our T shirts to us in that time. First of all, let's look inside of our file system, and we're going to look inside of the suppliers folder. And here we can see there's lots of different pieces of information from suppliers. And if we open up a few of them here, one from fabric First, another one from Kit Pro and another one from Print force, and then we have a little look inside. So here we've got a delivery note saying that fabric first, the wholesaler is our supplier. And on 28 January, we place this order with this order number. It's going to be delivered to us, courtesy of Kit Pro. And these are the products that are being delivered. So it's white, yellow, and blue and white striped t shirts, and that's going to be for our teams for World Cup T shirts, signed by myself. And if we look here inside of the Kick Pro delivery note, we can see some very similar thing here says what they're delivering the condition, but apart from that, pretty much the same. And then there's another one here, which is a delivery confirmation to say that on 18 March, a job with this number was delivered to the KIPro warehouse, and it was the Argentina fan T shirt, 300 of them printed and quality check. The result was it passed, which is great. And it's got some information about what they found and the quality here. The estimated delivery to us is 25 April, and the lead time should be five business days. So these are three different files here on our file system, which are the kind of things that we want to read and present in a dashboard in a different format. So we can see when these things are being delivered or if they have been delivered, or if there's any problems with our deliveries. So let's head back to Claude and see how we can generate a dashboard that can help us to see this information in a much clearer way. So here we are inside cow and I'm going to paste in a prompt that's going to help us to visualize all our information about deliveries from our suppliers a lot clearer in a dashboard. Here's our prompt, and we're saying read all files in passion sports suppliers. That's invoices, pricing agreements, and delivery notes, and produce an HTML supplier status dashboard. For each supplier, what we want to see is a status card with their name and current status color, where red is for overdue payments. Amber is for shipments due within seven days, and green is where we're in the clear. And we want to see the outstanding invoice total, the most recent delivery note summary, and lead time and next expected delivery. And then we want to save it to Passion Sports reports because it is a report in the form of a dashboard. So this is going to be a nice HTML dashboard that opens in any browser so we can share it with our operations team. And as usual, we want to see a plan for this first. So let's run it. So Claude has confirmed that it's seen everything in the nine files. It's said that it can see in kit Pro for this supplier that something's been paid, something's five days overdue. I can see what's outstanding, latest delivery times, next expected delivery, and the lead time three business days. And it's got similar information for all of our suppliers. So our other two suppliers here. So that confirms that it can see the files and it's got the information it needs. It's seen the total outstanding across all suppliers, and it's saying it's going to create an HTML page with a summary banner, and it will show the total outstanding number of suppliers at risk. Then three suppliers cards in a row. Each card will contain the supplier name colored status badge, so we can see the status. Is it outstanding? Is it overdue, or is everything fine? I asking, should it go ahead and build it? I'm happy with that, so let's do it. I've said yes, so run. So there you go. Our supplier dashboard we can check on a file system, and then we can look in the Reports folder and we can see it's being generated here. But remember, we can also always access the dashboard directly in here, either by clicking here or by clicking on this View suppler dashboard, HTML. They both do the same thing. You can also click here and just open it in Google Chrome. So here's our dashboard. If we zoom in a little bit, now have something really clear for our team. It says it's the supplier status dashboard for the Passion Sports W Weld cup 2026 launch, as of today, seventh of May for the operations team to view. There are three active suppliers, and the total outstanding is 9.7 K of product and three overdue invoices, which is actually all three. All three suppliers have overdue invoices. Fabric first have flagged a potential credit hold, so immediate action required. So this tells our team that it's something we need to act on quickly. So Kick Pro has 3.7 K outstanding the invoices it relates to. These are the products it relates to, so it's the Argentina fan T shirt for which 300 of them have been ordered and have been paid for. And then due 2 May, the Brazil T shirt, which is five days overdue. The latest delivery is 12th of April, and this is the reference for it. And the Brazil training T shirt, which is yellow, has 350 delivered. All items are accounted for. No defects. Lead time for the next delivery is three days, which is 28th of April, and these are the Morocco's T shirts, of which there are 200 and they're overdue. So the main issue here is that we have these t shirts, which are five days overdue. We know that there are no defects with them, but we just need them to be delivered. And we know that our next delivery is already overdue, and there's a three day lead time minimum on it. And we've got similar information for all of these for print force and for fabric first. So not looking too good, but at least the team will now know that we've got three deliveries that are overdue and are urgent. This one says that it's actually urgent. And the main reason is that it says that supply a flag that there's a credit hold risk if it's not settled immediately. So this needs to be paid for quickly, and it's five days overdue. So here we can see very quickly on this dashboard, it's much easier to see that there's a problem. Whereas, if you look at these other files, as you can see, it's much more difficult to look at this and immediately see there's a problem. Nothing's flagged anywhere. Nothing's in read. You'd really have to read all the information and work it out yourself. That's really helpful. The next scenario is it's end of April at Passion Sports, and we have a sales CSV file, a customer feedback file, and a supplier invoice summary, and we've got our own notes. And we need a written monthly report that we need to send to someone seen in organization. It needs to be structured and needs to be professional and under 600 words. So what we're going to do is we're going to take in these four files, and we're going to result in one formatted word document, and then CW can read all of these four files simultaneously, synthesize across them, and then follow our structure and save it as a doc file in Word format directly to our reports folder. We don't need to do anything, and that saves us time writing a report. So let's go over and have a look at the files and get into. Here we are in the file system, and we're going to go into reports April, and we're going to open all of these files up. So here you can see we've got various different documents. You've got some customer feedback from April 2026 that was collected from email replies, trust pilot Instagram DM. So this is all of the feedback we've got positive and negative in this file, sizing complaints, shipping queries, net promoter score, so how customers would either recommend not recommend or neutral. Looks like we've got some good recommendations there. We've also got so this is the customer feedback for April. We've also got some of my personal notes for the company for April about what went well and what needs fixing. And so, down the bottom, we've got a recommended actions for May, as well. And then we've also got this April sales document, which has a bunch of orders, what kind of products they are, and then their statuses and all the information about the orders. And then down here, we've got a summary of all the supplier invoices, what's been paid, what's paid, and what's overdue with their totals. So four completely different bits of information in four completely different files. And what we want to do is write a report in word format about this to someone in the organization, but we don't want to have to copy and paste bits out of all of these files. We don't want to even have to touch the file. We just want the report generated for us. So let's go into CW and see how we can do that. So here we are in cow, let's paste in our prompt and then get started creating our report. So the prompt says sing the following files in passion sports reports April, which is the April sales, customer feedback, supplier invoices summary, and my notes for April. I want cowork to write a monthly performance report for Passion sports. The structure should be in one report, I want to see the sales overview from the CSV, the top products, focus on the World Cup range. I want to see the customer feedback summary, the supplier cost, the key observations from my notes, and then one recommended action for May. The tone of the report should be clear, direct, business appropriate, and a maximum of 600 words. And then I want the report saved within the April folder in passion sports April report. And as usual, I want it to show me a plan first, so let's run it. So Cowors come back with a plan. It says the sales figures are here, 20 orders, 17 completed, two cancels, one returned. It summarized the revenue units sold and all the other information. It says it's got the suppliers costs. It knows the report structure, and it has confirmed that back to me. The format will be a Word doc in professional business report style with a maximum of 600 words, no unnecessary headers or bloat with clear sections and type pros and asking, shall it go ahead? I'm happy with so I'm going to say yes, and let's run it. So cow is completed. It says, six sections in this report have been validated, clean, and the report runs roughly 550 words, which is within our limit. It's given a summary of what's in there, but the easiest way to think about it is to actually look at it, so let's have a look. So back in our file system now within the April folder, we can see there's this new report here called Passionsports, April report dot dot. Let's open it up. And we can see that this report, if we zoom in a little bit, starts with the heading passion sports. It's April 2026 monthly performance report. It's the kind of thing that we could actually send through to anyone in the company. It's very clear and clean. It's got the different headings, six different headings for the six different areas. And it's for April 2026 prepared by myself. And what we're going to do is just quickly go over it so we can see the sections of what kind of data is in the report. So the sales overview, April, we generated one grand in revenue from 17 completed orders across 44 units sold. Two orders were canceled and one returned, which are excluded from revenue figures. Average order value was 62 pounds 44, so we're very early in the game. And here are the top products. England, Brazil, Morocco, Argentina, and other T shirts. We can see the amount sold, the revenue figures and the percentage of revenue along with the totals, which total up to 100% of this value. And there's a little summary saying that the England 2026 World Cup T shirt is the clear lead product, as we can see here, accounting for 47% of all revenue and 45% of units sold. There's the customer feedback summary. 12 customers responded to the April feedback request. Ten of the 12 would recommend passion sports, which is great. A strong result for a brand in early launch. Positive sentiment centers on the cotton quality and the fit. The issues required action, and it goes through what those issues were that came up that the customer suggested. Then we've got number four, the supplier costs, iruarTtal supplier spend was 29 grand. As of the end of April, three invoices remain unpaid, and here they are. The three invoices with the totals. Fabric first invoice was five K and was due second of May. And then key observations from working notes. So these are the notes that I knocked up. April review, England T shirt is outperforming all the others. Kit Pro is performing well operationally. Three day lead time is holding, and the April Brazil T shirt delivery arrived without defects. Instagram engagement increases, Brazil pre orders from resellers. You can get the idea of the kind of stuff that's in here. And in a recommended action for May is to settle the fabric invoice this week, everything else, which is the the Morocco blanks, the South Korea range, Phase two stock depends on maintaining the supplier relationship. So there's just information about what we should do within May. So the interesting thing here is that we've managed to compile all of this information from these documents. And if you look at these documents, you can see that simply looking at them, they're not as interesting to look at, and it would have taken a lot of thought to decide how to take the different bits of feedback and just word them within this report clearly. But Claude CO has done all that for us without touching a single file. We've got all the most important information taken out about the sales, about the supplier invoices. Customer feedback and my own personal notes all summarized into one report and all automated using AI without touching a single file. So in summary, in order to generate reports and documents in a dashboard or anything you need like this, you give CW multiple source files from the file system without even touching them, and it reads and synthesizes across all of them. You then define the output format. In our case, it was an HTML, dashboard, or a Word document, and then CW produces it. You can set a word count and a tone to keep the written outputs, usable or as clear as you want. And then CW saves the files directly to your specified path. No copying and pasting needed, making life a lot easier for us. Now it's your turn. What I want you to do is generate some kind of report like a monthly report, identify a report you write manually every month, every week, maybe even every day, in your own business or in your role, collect the source files that you take the information from. It could be CSVs, notes, feedback, invoices, anything on your file system, any of the files you have, and then ask cowork to draft the report from those files. Compare the time taken to actually do it using cow versus doing it the manual way. And I think you'll be very pleased. So have fun doing that and I will see you in the next lesson. 13. How To Schedule Tasks So Claude Runs Work On A Set Cadence: This lesson, we're going to go over how to schedule tasks, so Claude runs work on a set cadence. Scheduled tasks are one of Co Work's most powerful features. What you do is you define a task once. You set a cadence. So, for example, daily, weekly, every Monday at 8:00 A.M. Or whatever you want, and CoW runs it automatically. So in this lesson, we're going to show you how to set up your first schedule tasks, and we're going to use the running example of passion sports. So let's get into. So what we're covering today is the feature scheduled tasks. Why we use it to define a task once, and then cow can run it automatically on a set cadence. And what it solves the problem where you're consistently having to remember to run the same report every week or do the same data pool every morning or just essentially run the same task often and frequently. The scenario is you're during the production phase. We're during the production phase for creating our T shirts for the World Cup. And every morning at 8:00 A.M. We want a status briefing. So we want to know how many units are in production, how many are ready to ship, and whether anything needs attention before the day starts. And the goal is to set this up once, and then CW will read the files that it needs to do the job and save the briefing that we need every morning automatically without us triggering it. So we're fully automating it, almost like a worker who does the work for you at the right time every day and never forgets. So let's head over to our file system and see the kind of files we're working with, and then we'll go to CW and we'll set up our schedule. So here we are in our file system, and the two places we want to look is within this passion sports folder as usual, and it's within orders and then daily, so our daily orders. So we'll open that one up and here it is. And then the other one we want to look for is in suppliers and the supplier status notes. Open that one up. So if we zoom into both a little bit, on the left, what we have here is our daily orders, and you can see usual information, the order ID, date, product name, we've got the skew number which identifies the product, the quantity unit price, the total and the status, importantly, is it awaiting production in production? Is it ready to ship or is it at risk? So these are the important things to us as a business. And then we've got the customer email in case we need to contact the customer that this relates to. And let's go over and have a look at our other document. So this is the supplier status notes. And here, this gives us our usual overview of the statuses of our different suppliers. And so here, for example, for Kit Pro, the status is operational. Currently in production, we have a 180 units of the T shirts, and we have the English T and the Spanish T in production. The units are ready to ship. There's 95 them that are ready to ship. And it's the Argentina and the Brazil And issues there aren't any issues at the moment running on schedule. The lead time today is three days, and it's got similar information for the other suppliers. So what we want to do is build a report of these two pieces of information, which is something that we've done something similar for before. But the important thing here is we just want it to happen automatically every day at a set time. So let's go over to Claude and build that report right now, and we're going to automate that in our schedule. Here we are in CW, and the way to set a scheduled task surprise is to go up here to where it says scheduled. Click on that. And then here you'll see all of the scheduled tasks, and we can run tasks on a schedule or whenever we need them. And we can do it by typing slash schedule in any existing task to set it up. But what we're going to do is we're going to set it up within this area. So what we do is we go here and click New task, simple as that. We can give it a name. So in this case, we're going to call it World Cup Daily status Briefing. And in the description, we're just essentially going to say that this is going to be exactly what it says, a daily status briefing for the orders of T shirts for the World Cup. So I'm actually going to name this slightly differently. So I've renamed it World Cup T shirt orders daily status, and I'm going to write a little description that says exactly that. So I've said this is a status briefing with updates on the status of orders of our World Cup t shirts, nice and simple. So in here is effectively where we write the instructions for what Claude Cowork should do. And essentially, this is just a prompt. So I'm going to paste in the prompt that we created earlier. And here it is. And it says scheduled task, but that's not necessarily important. And it says that it runs daily at 8:00 A.M. But I'm going to remove this because we're going to set that up and actually configure it. This is useful if we run this scheduled task actually in the box at the bottom when we go back to cow. But for now, we're running it already in the scheduled task area, so we don't need to tell it that now. So I'm going to remove that. So now, it says, read the latest files in Passion sports orders daily and passion sports suppliers. Produce a morning production briefing save to passion sports reports daily, file name, production brief, with the date in the form, month day, and that's going to be a text file. And in this briefing, it needs to include the units ordered but not yet in production, the units in production per supplier, the units that are ready to ship, any supplier issues flagged in the latest emails or notes. Any orders at risk of missing the delivery deadline. Keep it under 200 words factual and no padding. Down here, we can say work in a project. I could say passion sports, but because I've already told it the full path, I'm not going to do that again in case it thinks that there's a subfolder called passion Sports. So I'll just let it figure it out based on this full path. Here you can choose whether Claude should actually ask you before acting or act without asking. And because it's a report, I'm going to tell it to act without asking because we want this to continue. Even if we're not at the desk, maybe we leave our computer on, and it will act, even if we're not at our desk. Now, what you've seen pop up here is it says, Claude will work without pausing for approval. This can put your data at risk. And it says, Claude will use your connectors without pausing for approval. So I've said it can act without asking. It won't need to use any connectors, but if it needs to, it will. I just happen to know that it won't I'm happy for it to act without asking. I've also only given it access to my business accounts anyway, so it's not like it can look at my personal information. Just be really clear and really cautious before you allow Claude to act without asking, but know that it's the best way for it to run automated tasks without you needing to be at your desk and approve anything. Over here, we can pick which model runs. Now, this is a pretty simple task. So Sonic 4.5 is called because it's normal processing. I don't want it to use, and more expensive one, it doesn't need a lot of reasoning here. It's pretty straightforward. I'm going to leave it on Sonic 4.6. And the frequency, if you leave it on manual, then this means that you can run it whenever you want. However, we want this to happen daily, so I'm going to pick daily. And we said we wanted it at 8:00 A.M. So let's just update that. So there you go. This is our schedule task all created, and then all we need to do is click the save button and that will be saved as a scheduled task. So now you can see over here, it says World Cup, T shirt, orders, daily. This is our schedule task here, and you can see the description. And down here, it says that it will run every day at 8:00 A.M. If we double click on it, you can see it opens again, and it's got the title here, description. It's active because it's been activated. And the next run is tomorrow at 8:00 A.M. This will run tomorrow at 8:00 A.M. However, it's not 8:00 A.M. And I want to run it. So what I'll do to demonstrate it is I'll click the Run now button and you'll see this, do exactly what we've asked it to do. So here it is, and we can always click the FE button to see what it's doing. You can see Claude's working on it and thinking about what it's got to do. Here you can see it runs today because I've just activated it. It's telling us the progress, and it runs just like it would if we had pasted the command here in the box. It warns us also that Act without asking is on, so it will use any connectors it it needs to and browse the web if it needs to. And there you go. It's just finished. So the automated task is complete with a flag for my attention. It said the task ran, but the source data directories could not be found. So it turns out that the reason why this happened is because we actually needed to specify the folder that these sit within. So if you go back to the file system, we can see that here, if we go to documents, we can see that Passion Sports actually sits within documents, but we didn't explicit give Claude Co Work access to that. So what I did was I went back if we go back here, and I went to the schedule task, then I went inside of the schedule task. And if we edit, then go down to the bottom here, we can see here, I've now given it access to the documents folder, and that is where passion sports sits. So give it access to the folder above the one that you mentioned, and the one that we mentioned was actually Passion sports. So it can work within so if you save that and then run it again, then you'll end up with access to the right folder. So I've run the task again, and it's run successfully this time, which is great. And so it says Production brief for ninth of May 2026 has been written to this folder, and it summarize what's been captured. But we're going to see a much clearer picture of what's been captured if we just open it up, so let's do that. So inside documents, passion sports folder inside reports, and then daily, we can see that the production brief has been written, so let's open that up. And if we zoom in a little bit, we can see the files successfully been written, we can see that there are three units awaiting production and what they are. The units in production for each supplier, and we can see here for each supplier what's in production, the units ready to ship 95 units and what they are, any supplier issues and orders at risk, all the information we asked for. But the important thing is not this document, it's the fact that now we've got a way to schedule any task we want within Claude by just going here and creating a new task, and that has worked absolutely perfectly. So the way to schedule a task is go to scheduled new task, paste in the brief, which is essentially your prompt, your instructions of what you want to do, set the cadence to daily and set the time to 8:00. That's what we did, but set it to whichever cadence and to whatever time you click More Options, select the right model. And if it's something really basic, I've been using Sont, but you can select what's appropriate for you. And then, importantly, select the cow work folder that's going to run in relative to what you're saying in the instructions, save it, and then you can either run it to test it, or you can wait for that time, and it will run itself. So the important thing to remember is that scheduled tasks only run while your computer is awake and the claw desktop app is open. So in summary, you can use cowork to start a new scheduled task, write your brief, set your cadence, and save it, and then it will be scheduled to run at that time. Cadences are usually hourly, daily, weekly, or weekdays. You can actually slash schedule in any task to trigger the setup flow. In other words, if you start a new task and you type in slash schedule, then you can give it instructions, and it will actually set up the schedule for you. And missed runs actually execute automatically when the app reopens, so that's good news. The key thing to remember is if you schedule a task and you want it to run on time, your computer must be awake and the clawed desktop must be open for the schedule task to run. So now it's your turn. What I want you to do is set up your first schedule task. So identify one task that you trigger manually every day or every week or periodically in your own work. Then set it up as a schedule task in coop, write the brief for it, which are the prompts or the instructions for it, set the cadence, and save it. And then let it run on its own and then check the output whenever it runs, check that it's exactly what you wanted, and I'm sure you'll be very happy that you now got the ability to schedule task, almost like your own little worker doing work for you in the background. So I have fun doing that and I will see you in the next lesson. 14. How To Set Up Recurring Workflows And Walk Away: In this lesson, we're going to go over how to set up reoccurring workflows and walk away while they just perform for you. A scheduled task runs one thing on a cadence. A reoccurring workflow chains multiple steps into a single schedule task. So Cowork runs an entire sequence of work automatically. And this lesson shows you how to build a complete recurring marketing workflow for passion sports. So let's get into it. What we covering today? We're covering recurring multi-step workflows. Why we use them is to chain multiple task steps into a single recurring workflow. So for any role, any job, anything that you do, there are usually a set of tasks that need to be done to achieve a particular outcome. So if you're a salesperson, you may need to have a lead acquired, and then after you get that lead, you may need to send them a follow up message, then get the sale, then book them in for a follow up meeting. That is a workflow. And as a developer, sometimes you have things in your to do list you need to do, then you move them into progress, you write some code, and you're done. In this case, what we're doing is we're solving the problem where we need to run multiple steps for something in our business. So what it solves, in this case is running each part of a weekly routine or a weekly set of reports or anything we need to create separately, but we can do it one trigger at a time, and that one trigger will initiate a whole set of steps, which we call a workflow. So let's get into it. So the scenario is that every Friday at 5:00 P.M. Cowork is going to read the weekly sales data and identify the best performing products based on that data and then produce two customer facing assets. HTML email newsletter and then a product launch page. And these are ready for our team to review on Monday morning. So we don't need to put them out directly, so we'll always have a human in the loop, but it automates the process for us, and there should be very little or in most cases, nothing for us to do. So that's what we need to get done. The outputs will be a branded HTML email with our usual T shirt image in SVG format as a hero visual with the headline, the copy, and the call to action, CTAs call to action to ask the customer to actually buy the product. And in the product launch page, we'll have the T shirt visual, the name, the price, the sizes, and then add to bag button. And both of these, both the email and the product launch page will open in any browser, so we can add the product launch page to a website, and the email we can add it to our email marketing system so I can get sent out whenever we want. And we're going to get Cowork and AI to do all of that for us, create those two pieces for us. Let's have a quick look at the files that are going to generate this for us that Cowork is going to use. So here we are in the file system, and inside the Passion Sports folder and the reports, here's the weekly report with our sales report, and this tells us how well the sales are doing for each product. So if we open that up, we zoom in a little bit here, we can see here that we've got all of our different products and we can see how many of each is being sold. So in week 18, we can see the best selling one is the England World Cup T shirt in white, and we can see 47 units sold and the revenue. But the main thing is we can tell which is the best selling product at this time. And now, the next file we're going to use if we go over to the World Cup 2026 or actually the products World Cup 2026 folder, we can see that there are product files, spec files for every single product as usual. So if we were to open up the English England T shirt spec file and zoom in a little bit on that, we can see here that it's got all the information about that t shirt. So what we can do is we can use these two bits of information, the best selling t shirt at this time, and then all the information about each T shirt, for example, if we just want to focus on the best selling one, we know the best selling T shirts price, what territory it sold in, how much it cost, the colors, and all of this good information. So we can build anything we want. Based on knowing the best selling t shirt and all its information. So let's go over to Cowork and then we're going to work out what our workflow is. What is it that our team does with this information every single week? So here we are at Cowork and we can past it in our prompt, which will show us what the workflow is for our team. So here we are. This is our recurring workflow. Now, if you can imagine an assistant or somebody who works in the marketing sales team or someone who just assists one of the managers in the company needs to run this every single day, and now we've given them less to do. They can still be in control of it, but they can organize running it. So let's go through it. So this is a recurring workflow that runs every Friday at 5:00 P.M. And step one is to read the Passion Sports reports weekly folder and find this week's top product by a unit sold from the weekly sales CSV that's within the folder. So that's what we just looked at. So step two is to read the Spec file for that product from Passion Sports Products Well cup 2026. And that's the other file we looked at. So if it finds the top product is the English T shirt, it will find the Spec file for English T shirts in here. Step three is to produce a customer facing HTML email newsletter. It should have the SVG, which is an image image format, a flat lay of the T shirt as the hero visual, and then a bold headline with the product name and then World Cup 2026, and then two lines of fan first body copy and then a quarter action button which says, shop now at passionsports dot code at UK. And then a footer with all the social links. So this will produce a nice email newsletter that can go out to our customers, and then it should save it in Passion sports marketing launch email of the right year and week. And then step four is to produce a launch HTML page. So a full product view with the same T shirt, graphic, a name, price, sizes, description, and add to bag button that allows a user to add that to their bag, a potential customer to add that to their bag. And save that HTML file in Passion sports market in product page with the year and the week, and that's an HTML page we can add to our website. Then it says, If any step fails, log the error and continue. Don't ask for approval, run all steps automatically. So what this gives us really is a workflow because it's a set of steps that a human would usually do, and we can completely outsource this to Cowork. So the only thing we need to do, as we found earlier, is this says Passion Sports folder, but Cowork won't know which folder to look in to find it. So what we do is we go down here and work in a project and we say go into documents, allow Claude to change files in documents, allow and now that I've allowed it, it will know that it's relative to the documents folder will look for Passion Sports folder. And so now that it's got the recurring workflow here, and it runs every Friday at five, we're going to see if Cowork will understand to schedule that task for us. So let's go. So here we can see that Cowork actually did the work. It we know that it hasn't created this as a scheduled task yet, because if we go into scheduled, we can see all of the scheduled tasks and it hasn't created a new one. But this was just a test for me to see what it would do. I'll show you how we can get it to actually create it as a scheduled task. So if we go back to what we've done, Claude Cowork has said First of all, it had all the data and it knew what to build. Up here, you can see it told us our progress. It was crossing them off as it did the work one by one. So I knew it wasn't scheduling. It was just actually doing the task. And that was fine, we could have stopped it at any time. But I let it run. So here you can see it's telling us the steps it went through. The workflow was complete with no error. It identified the top product as the England 2026 world cup T shirt, white, so that's correct. S two, it loaded the spec, particularly for the English T shirt, which is great. And this is where it got all the information about English T shirt. Step three, it created the email newsletter, which we're about to go and look at, and step four, it loaded it created the product launch page, which is an HTML page. Both of these HTML pages. So let's go into our passion sports marketing folder where we told it to create these pages and have a look at what's generated. So here in the file system, we'll go to Passion Sports and then marketing, and we can see we've got a couple of new files here. This is the lookbook we generated in a previous lesson. But first of all, we've got the email which is going to go to our user. So let's open that up, that'll open in Chrome. So, great. Here we go. So we have now got I'm going to zoom in a little bit here, but we've now got just a basic email. It says Passion sports World Cup, 2026 collection. We can always jazz this up with logos and things. But right now it just shows a very basic image of the T shirt, and then it says, Week 18 hero Product World Cup 2026, England, 2026 World Cup T World Cup, 2026. This is the T the terraces demanded clean white with a bold navy graphic that does the talking before a word is spoken, crafted from 100% stands for global organic textile standards. So this is the standard for certifying this as organic cotton. So this is certified organic cotton. It's fan first in every thread. Wear it to the match, the pub or wherever the tournament takes you. And obviously, this is really basic. We haven't done any work to get the design of the actual email or the copy to be absolute best yet. But for now, this is a great example of what it can do. We've got the email. We've got a button which says shot now at passionsports dot code at UK, we can click on where the user can buy it, and we've got links to the socials where they can get more information about it, copyright, and everything. And this is just from a really simple command, so I'm happy with that. Let's go and look at our next page, which is going to be our webpage. So back on the file system, we'll go to this product page, and let's double click and open up that incro. We go. Here's our website page. It's given us the passion sports where we can replace a logo here. We've got some temporary links here, which go to the men's, the women's, anything to do with the World Cup. And there's a sale page. So this is obviously a website that's been created for us by Claude, so it's just adding what it thinks should be in here. So I'm going to zoom in a little bit. And you can see we're in the home World Cup 2026. England 2026 World Cup T shirt section. It says, This is the week's top seller. It's got a SVG image, and it's showing here the unit number, the price, free delivery in the UK over 40 pounds, including VAT. So Claude does a little bit of improvisation to make this look realistic. And it says, The T the terraces demanded a clean white base meets a bold Navy chess graphic with a red accented color that wears its colors with pride without borrowing them from crafted from 100%, and this is, again, the certification body certified organic cotton, 180 GSM. It's fan first in every thread. Generous regular UniSEx fit works from the stands to the street. It says that the material here, 100% cotton, the certification, the fit, regular unisex, colors, design, the print. And then it's allowing you to select the size here. So this is a real webpage that we could actually use and implement. You can select the quantity and you can add to your bag. It says added right there. As in Stock, 145 units available, ship in one to two working days, secure checkout, free UK returns, ships within 48 hours. So this is a pretty realistic page. We can always enhance it and add things, but it saves us getting a web page builder just to create the basic version of it, we can get a proof of concept. And then down the bottom, it's got links to our socials. So I'm happy that it's done exactly what we asked it to do. So now back in Cowork, once we're happy that this actually works, we can create a scheduled task for this by as we did before going into scheduled and creating a new task. But another way to do it is we can do it directly from Cowork from the chat box in Cowork by going back here and simply asking Cowork to create a scheduled task for this and then making any updates if we need to. So I'm going to do it that way. So, here it is. I've said, Great. Create a schedule task for this that runs as specified and on the cadence specified in the above command. And I said that because in here, we already said runs every Friday at 5:00 P.M. So let's get that started now. So we can see it's creating the schedule task. So here now it's popped up this window, which says, Schedule task Friday passion weekly launch every Friday at 5:00 P.M. Find the top selling World Cup tea generate launch email and product page HTML at 5:00 P.M. Only on Friday. If you want to, you can see more details here. So this essentially creates a simplified version of the text that I gave above or really something that is more understandable to Claude. So there may be some tweaks to the wording, but it amounts to the same thing. So, for example, it says, You are running the Passion Sports weekly Friday marketing workflow, execute all steps automatically without asking for approval, et cetera. So it's worded very slightly differently to what we said here where I just said reoccurring workflow runs every Friday at five. But it does all the same things. It's got all the same steps. And also, what I like is that Claude puts in details here to make sure that it stays on track, based on what we've generated already. So, for example, it says here, a dark navy footer with SVG social icons. I didn't say that in my original instruction, but because I've approved what it's done, it's added all of that so when you review and you're happy with it, you just click this schedule button, and then you go summarized for us that it's done. The scheduled task. Friday Passion weekly launch is live and will run automatically every Friday at 5:00 P.M. I will do all the things that I've asked it to do, and I'll get a notification here when the run completes. You can also manage it from the scheduled section on the sidebar. So the good thing is now, if we go over to scheduled section, even though we didn't go in here to configure it, we can see it's here. So if we click on Scheduled, then have a look down here, and if we double click on this, we'll see this is the Friday Passion weekly launch. And everything that we just read is actually detailed in here. You can see that it repeats every Friday at five. You can also see that it's always allowed because I asked it in the instructions. The only thing it does do it gives us a warning that this task runs during peak hours weekdays five to 11:00 P.M. And then 1:00 P.M. To 7:00 P.M. Local time, and we'll consume my session usage limits faster. So if you don't want it to do that yet and you're not ready, you can always deactivate it. If you're ready for it to run now, you can run now as usual. But you can see that's totally scheduled, and it will always be here. So if you want to reactivate it, you can always double click inside and reactivate it or just run it manually. So that's great. I'm really happy with. So a key principle here is regarding the do not ask for approval, command or setting. So use this for reoccurring workflows that you trust and include this in your brief because it tells cow to run all the steps without pausing for confirmation. Especially if you want this to run like a little agent or a little worker, you usually want it to run that way once you've tested it and you trust it and you're not going to harm any information on or off of your own computer. So only use this after testing the workflow as we did manually, at least once. And usually you would say, show me the plan first. And also, before you schedule it, as I've said, yes, test any recurring workflow manually, at least once with the show me the plan before scheduling it. But also, remember in our case, the HML email and product page are drafts for review. We always like to have a human in the loop, someone who can visually check that what's going out to the public is acceptable. There's no sensitive information, and essentially, it is what we want it. We may even want to replace add some images for the branding or anything like that or update it and ask it to add some images for branding. And essentially the team are the people who actually published them. We haven't published them live yet. So Cowork doesn't send or publish them automatically, unless, of course, we were asking it to, which we haven't in summary Cowork allows us to chain multiple steps into a single reoccurring workflow brief. You can schedule it via the scheduled menu option on the side and then create a new task, set your cadence and save, or you can do what we did and just tell it within the chat box in Cowork to actually schedule it for you, and it will allow you to approve that. Include the command to do not ask for approval only after you've tested it after you've tested the workflow manually to make sure, and be sure to add any error handling that you need so that if any steps fail, it will log that and continue because oftentimes we're not at our desk when this is running, so it will actually log to our file system what it's done, and we can always go back and investigate that. Now it's your turn. What I want you to do is build yourself a reoccurring workflow. Take a weekly routine from your own work that involves more than one step, test it manually in cow once with Show Me the plan first, and then set it up as a reoccurring scheduled workflow that you can walk away and be really confident you'll come back, and it's all done for you. So have fun doing that and I will see you in the next lesson. 15. How To Use Claude In Chrome To Browse And Research On Your Behalf: In this lesson, we're going to talk about how to use Claude in Google Chrome to browse and research on your behalf. Claude In Chrome brings Claude directly into your browser. It can read the page that you're on, extract data and save findings to your local files all without switching Windows. And it can do that even without you going to the browser, and I'm going to show you how to do all that. For passion sports, this means researching competitors and supplier pricing and gathering market data absolutely automatically. So let's get in. What we're covering today is clawed in Google Chrome. Why we use it, we use it to use Cloud inside of any websites to read pages, extract data, and save the data to local files or wherever we want. And what it solves is context switching. So often when we're working, we want to stay focused on a task, and we don't want to have to switch between browsers and then clawed and then the file system and then manually copying web content into files. We want to be able to automate all of that. So let's get started. I'm going to show you the files that we're going to be going to and how we're going to use Cloud in Chrome. And here we have the product range, the usual unit number, unit identifier, the skew, the product name, territory, UK, the colors, recommended retail price, the sizes, the stock, and the status. And what we can do is we can look at our current product range and we can compare what we're doing for all of these things, sizes, colors, retail price, all of this stuff. We can compare this to what our competitors are doing, and then we can either make some decisions or generate something based on the comparison of what our competitors are doing and what we're actually doing. So let's head over to Claude and work out how we can do information, how we can use laud for Chrome to help us do some research based on the information we have for our products and then look at our competitors products. So the first thing we need to do when we're at cow is we need to make sure that Claude for Chrome is installed on our browser on our Chrome browser. So here's our Chrome browser, and we know that Claude is not installed because there would usually be a symbol here, and what we want it to do is we want it to be installed as an extension in here and we don't have any extensions at all. So if we head back to Cork, what we need to do to install it, the best way to find the install button is to, first of all, go over here and click on the initial button, then we want to go into settings and then go down to connectors, scroll down a little bit, and you'll see it says Claude in Chrome included, and then we can just go into configure. And in here, we can see that it's actually disabled at the moment. It says Cloud in Chrome. Let's Claude handle work in the browser via clawed desktop. And that's exactly what we want. So we click Enabled here, and then we click this install button. Puzzle piece will appear in here. And if you click in there, you can see Claude has now appeared, which means it's been installed. But we can do two things. First of all, we can pin it, and that means that we will get this symbol up here. And then once it's pinned, there are actually two ways to access Claude to work with your browser. The first is actually within Chrome, and the other is within Claude cow itself or within Claude chat, in fact. So I'm going to show you the first way. So if we click this button here, you'll see that a panel comes up on the side. It warns you that this is a Beta feature and to be careful, essentially. So I'm going to click. I understand. Then it goes through a little bit of introduction to what it can do. It can automate your repetitive task, which is what we want to do. It's going to have tab group access, which means if it's running in your browser, you'll actually see open tabs as you can see here, and you can see Claude is managing this because it says Claude and it's highlighted them. So we'll click next on that. And it says we can use shortcuts to save time. So what will happen is when Claude is running, sometimes you'll see that there will be some shortcuts created that look like this for particular tasks that you set up. So I'm going to say, let's go. So now we can get started. This looks a lot like the chat windows that you're used to, but we're now actually running Claude within the browser. So it's got a lot of the same features. We can set the model here, and down here, we can tell it to ask before acting or we can tell it to act without asking. When we give it a task, it can just run through every single piece of that task and deliver for us. But I'm going to leave it and ask before acting for now. We can also drag out this window if we need a bit more space. So here I am at alibaba.com, and for anyone who works with ecommerce to buy products that they can sell on the market, you'll know that alibaba.com has been one of the main places that many sellers go to buy in bulk. So what I've done here is I've gone enough search for the best selling World Cup T shirts, and it's given me this list of t shirts. So let's go now into Chrome and do a little bit of research on these t shirts. So I've put in a command that says, Extract the pricing on this page, format it as a CSB, separated values with columns, product, minimum order quantity, which is the minimum amount of orders that I need to place to buy these products. So here you can see it says the minimum order is two pieces. Was here, we have to buy 500 pieces, price per unit, lead time, and save to passion, sports, suppliers, and that should say Alibaba prices. So let me just update. And since I'm running it here, I've actually updated this to say download as Alibaba prices and this date dot CSV. So what that now should do is look through everything that's on this page and get me a price list. And that will help us to work out as these are some of the popular T shirts here to work out what the prices are from this supplier or from these suppliers, and then make some decisions about how many we can order and what the price is likely to be. So let's run that. So, again, I get the plan here which says to read the current Alibaba search results page, identify pricing details including the details I've given it, format the extracted data as a CSV, and then download. As this file name. So I'm going to approve that. So within a few minutes, it tells me that the CSV Alibaba Prices has been downloaded successfully. It gives me a summary of what's extracted, but it would be easier to just download it and have a look. So let's do that. I'm going here into my downloads fold, I'm going to open this file up. And here looking at the file, you can see we've got a long list of all the products here. We can see, for example, this Taunton sports custom vintage retro football jersey, soccer Jersey T shirt for men, women, kids, street style. They often have really long titles in Alibaba, but the minimum order quantity is two pieces, and it sells for 662 to 929. The lead time isn't available. So what we can see is that some of these t shirts, I need to buy a minimum order of two pieces, whereas others, I need a minimum order of 500 pieces. Also, you can see if I spend more, if I buy more, I get cheaper price per unit. So this is the kind of information that's really helpful. We can also get more specific on what we're searching for to exclude some of these. We may not be interested in kids products. But as we can see, it's generated that for us, which is pretty helpful. So I've shown you what we can do within the browser, but the important thing is we want to use cow, and we already know that we can read and write from our file system, our file structure really easily. You saw we had some issues, and it was taking quite a long time. And in the end, I just had to tell the browser to download the files. So let's go over to CW and do another piece of work and compare what we can do there. So here we are at CoWork. I have created a new task, and let's paste in our instructions. So here are the steps of what I want to do in our workflow for research now, and this is going to be using the Claude Chrome browser, but now from CWO. So step one is use Chrome to visit this website, which is google.com slash TENS and search for World Cup 2026 merchandise. Note the top related topics and any rising searches. Then step two is to read the passion sports products, current range 2026 XLSX. So that's the spreadsheet we showed earlier with our current range. Set three is to compare the trending terms to our product range. And we want to answer the questions, which trends match products we already carry? Which are gaps we could target before the tournament. So this gives us some research really that says, Here's what we're currently selling. Here's the trends that we're currently seeing, and then we're going to get cow to compare, are we already on trend or where do we need to improve before we go to market? So, the only thing I'm going to do here is I'm going to pick from this list to make sure that it knows we're working within documents, and I'm going to allow that then now all I need to do is run it, and it's going to run through this to make sure that we can do our trend analysis. So let's get started. So Claude says that it's going to hit Google Trends. If we go over to our browser, we can see it's opened up this page, and this is the group that I told you about earlier. So it's opened up a tab group. Within there, search for World Cup 2026 merchandise, and it can see this page. And down the bottom, you can see, these are the rising trends. And if you flip this, you can actually see top trends as well. So it's flip the switch to see the rising trends, and I presume it looked at the top trends already. I can see that it's looking for these keywords within the United Kingdom in the past 12 months. And this is showing the graph of interest over time. So it's been sporadic. It hasn't been continuous. And then down here, as I say, it's got rising queries, so you can see people have been searching for the merchandising topic FIFA World Cup. And here, related queries are FIF Worlup merchandise, FIFA World Cup 2026 merchandise. So this is what this search has brought back, and this is what is going to be used within cowork to compare to our t shirts and the words we're using within our t shirts. You can see also here that on the next page, it's a little bit more specific. So people have been searching for Scotland UK constituent country, so that indicates Scotland is quite popular, people have been searching for shirt, garment, 2020 forms of cricket, so some stuff not related to football. So essentially, Cowort will do what it thinks it needs to do on this page to find out what's trending and how that compares to what we're doing already. So here we are back at cow, and it said, Good data coming through, let me grab the next page of topics and switch the top queries. So that shows, as I was showing you, it was going through some of the pages there. And it says Topic six to ten show, shirt garment breaking out, which is very relevant. Now let me switch related queries to top view and get the next topic pages. So this is where I was showing you that it was also showing related queries. And then it says, I have all the data I need. Let me mark the trends task complete and move into analysis and document creation. So it's now creating the file and writing the document. We'll wait for that to finish and see what it comes up with. You can also see progress here where it search through Google Trends for World Cup 2026 merchandise. It read the current range 2026 spreadsheet, which is our current range, and it's now compared the trends to the product range to identify gaps, and now it's going to write our document for us. So Google Trends document has completed. It says that search interest peaked at Index 100 in the week 12 of April 2026. The pre tournament buying surge is at its highest right now. Scotland was the number one UK region by interest followed by England, which is interesting. But we actually have the report. So as usual, we can access it here, and we can also check on the file system where we told it to write this. So passion sports marketing trend analysis. Let's have a look there and make sure it wrote that. So let's go to Passion Sports Marketing and here we can see the trend analysis document has appeared. So if we open that up, and if we zoom in a little bit, for some reason, it's giving us a blue background here. So I've just changed the text color to make it a little bit clearer. So this says essentially that this report analysis analyzes Google Trends data for World Cup 2026 merchandise in the UK over the past 12 months and maps those findings against Passion sports current product range, and the tournament kicks off in June 2026 across the United States, Canada, Mexico. Such interest reached its peak index score of 100 during the week of 12 April 2026, signaling that pre tournament consumer demand is now at its highest point. So that's not too long ago, so that's pretty good for us. Several strong matches with our existing range have been identified, so good news all around. So if we have a quick look through the Google Trends findings, it says that the search term World Cup 2026 merchandise showed minimal activity through summer and autumn, 2025 before accelerating sharply from January 2026 onward. So that means we've either peaked or on our way up at the moment. So if I just skip through this, it says, Win the United Kingdom, Scotland recorded the highest relative interest. So that means focusing on our Scottish T shirts might be a really good thing rather than just England t shirts. And it says related topics arising that are all breaking out at the moment. It says merchandising FIFA World Cup and FIFA World Cup 2026. So that means all of these topics are breaking out at the moment, and they're in very high demand. People are searching for these, and Scotland pops up here. The related queries are FIFA World Cup, which indicates that if we were to put the word FIFA World Cup somewhere in our clothing, that would help. However, it's probably a copyright issue. So we'd probably need to speak to FIFA and make sure we obtain permission or we license the right to use their name. However, it's a good option to have, and it's good to know that people are searching specifically for FIFA rather than just World Cup. It then shows our current range. And you can see, if we look down here, in terms of what's live, we don't have a Scottish T shirt live, and in fact, we don't have a Scotland T shirt anywhere in this list. So this information is really helpful because it tells us if we were to focus on one, we may get more searches on Google. However, what it does say is that based on these current searches, England interest is a top UK matching product here. England 2026 World Cup T is definitely a matching product for this trending search term, which is good for us. So here, it confirms some of the things we said that the GAP is a Scotland T shirt, which we don't have, and it tells us why, it tells us the evident, and it says the priority is critical. And it says the actual host nations t shirts, as well, are good ones to focus on. So we don't shirts for the USA and Mexico, and it tells us more information of that nature. Then the recommended next steps here are to immediately develop a Scotland fan t shirt and aim to get it live in the next two to three weeks, reactivate Germany and France products as well, and then commission host nation T shirts for USA Canada and Mexico, and there are some other suggestions as well. So all in all, this report is really useful. There's more detail we could go into if we take time to read this. But as you can see, just from running some research against the Internet using our Claude in Chrome browser extension, we've got some really timely information that's going to help us in our work and in our business. So in summary, in order to enable Claude in Chrome, enable it in settings, connectors, find clouding Chrome, toggle it on and then install the extension, remember to pin the extension to your Chrome tool bar for easy access. Claudine Chrome reads pages, extracts data, and saves to your Cork folders. And if you trigger it through cow, you'll find that it's much easier to write from write to your folder, as you saw. It works with Google Chrome only, not brave RC or other Chromium browsers at this time. And you can combine with cow task for web research and local file work all in one workflow. So now it's your turn. What I want you to do is use Claudine Chrome for real in your work. So I want you to install Claudine Chrome if you haven't already, go to settings, Connectors, toggle on and then install, and then open a website relevant to your own work. Could be a competitor, supplier, a news source. It depends what you're doing, but find something relevant to your own work. And then ask Claude Ichrom to summarize or extract something useful for you from that page and save it to a folder on your machine. You'll be happy that you did, and you'll see how much of a time saver is so that you don't have to do it yourself. So I have fun doing that, and I'll see you in the next lesson. 16. How To Use Connectors To Link Cowork To External Services: In this lesson, we're going to talk about how to use connectors to link Cowork to external services. Connectors link CWT directly to the tools you already use, like Gmail, Google Drive, Google Calendar, Slack, Notion, Microsoft 365, and at this time, at least 38 plus others and growing. Once connected, Cowork can search, read, and act across them from a single task without you switching apps. And I'm going to show you exactly how to use multiple different apps directly from Cowork. So let's get into it. So what we're covering today is connectors. Why we use them is to link Cowork to external services so that Cowork can read and act across your other apps. And what it solves manually copying information between Claude and your tools or doing research across separate apps. So let's get started. I'm going to show you exactly how we can do that. So here we are at Cowork and we're about to connect some different apps we call connectors into Cowork so we can start using them. The way to do that is to go down here to settings as usual, so we can go into settings. And then if you look by the side here you'll see it says connectors, so click on that. And then it shows you here that connectors has actually moved on this occasion to customize. So you used to be able to get to it from two different places, and in a way, you can. But right now what they've done is they've moved it to customize. So I could click here and it will show you, but I'm going to show you where customize is. Depending on the version you've got, it may still show connectors here, which is why I've showed you. So if you now come outside of this and we go to customize, you can see connectors is shown here. If we click on that, then you've got a list here of all the different apps that we actually have connections to. At the moment, I've got active campaign and Gmail, but for this lesson, I'm going to need a couple more. Well as GMO, which we do need, I'm going to click the plus sign here, and then I'm going to click Browse Connectors. And then I'm going to search for the connectors as it happens, they're already here at the top. So we've already got Gmail connected. We now also want Google Drive, so I'm going to click Plus on that. And at this point, it's going to get me to log in to Google Drive. So it's automatically connected, and I'm going to choose this account, click Continue, and it's going to say, make sure I trust Google Drive and say continue with that. It now wants to open a desktop app, so I'm going to say allow Claude AI to open it and open Claude. And it says, taking you back to the desktop app, you can close this tab. So I'm just going to close the tab, and then I'm going to go back to the desktop at myself. So you can see now that Google Drive is connected in here, and it says all the things that it can do here, such as download file content, get file, metadata, et cetera. So these are basically all the permissions that it has. And it's showing that it's essentially showing that these read only tools are set to needs approval, so it needs approval to read, and then the write or delete tools also need approval. So I'm going to leave it at that for now. So let's add our next connectors, so if we click the plus sign again, Browse connectors. And I want Google Calendar because we've got some stuff we need to read from our calendar, so I'll click the plus sign again. Again, I'm going to log in with this account, and I'm going to confirm that. Then we're going to decide what Claude should have access to for Google Calendar. So I'm going to allow it to view events, see and download any calendar, and I'm going to also allow it to edit because I think I need these in the future. Even though at the moment, I only really need to view events. I'm going to give you access now. So click Continue there. And the same thing, we can just close this tab and then go back to our desktop app. So now we can see the apps that we need are all in there. We've got Gmail, Google Calendar, and Google Drive, all connected. And that means when we run any tasks, it will have access to these with the permissions that I've given it. And as you can see, it says, as usual, our Read only tools and write or delete tools, they all need approval before they can do anything, and that's the same for everything that I've added. So now let's go back to Cowork and get crack in doing some work. So before we do any work in Cowork, let's go through the scenarios that we're going to do work for. So the first workflow we need to do something with using connectors is where Cowork actually reads our actual diary. So what happens is that Cowork connects to your Google Calendar. It reads the World Cup match dates that we've already added to the calendar, and it pairs each England, Brazil and Argentina fixture with the right passion t shirt, and then it drafts a match day social post time for 2 hours before kickoff. So based on the calendar, we'll know exactly what's going on in terms of matches, and then it can pick the right t shirts to promote on socials as well as the right kind of social posts to write. Why this matters? It matters because Cowork sees your actual schedule and then it acts on it. It's not just a spreadsheet that you posted, but it's the calendar you've already set up and use every day. And we could actually use something live on FIFA's website. This time, we've decided to just use a calendar because it can be a lot more useful. We can change things on the calendar. Let's have a look at the actual calendar that we're reading from, and then we'll go over to Cowork and get it to read it. So here we are at the Google Calendar, and you can see there are a number of different events on my calendar. So let's just open this one. This is a World Cup event on 15 June. So if I open that up, you can see that we've got the date, 15th of June, 3:00 P.M. To 5:00 P.M. We've got the location of the SoFi Stadium in Los Angeles in the USA. And you can see down here, it says, England's first group stage match passion T to feature, England, 2026 World Cup T why post 2 hours before kickoff, it's got some information there that's going to help us to do our posts. But this is an example of what we've got as a calendar item in our diary. And up here, it says England versus Group B opponent, World cup, 2026 group stage, so we know that it's to do with the World Cup. So if I close that out and just to give you another example. So that was on the 15th. So just to give you another one on the 18th, we've got Brazil versus the group F opponent, World Cup 2026 group stage. And that is at the Metlife stadium. If we zoom in a little bit, the Metlife Stadium in New York in the USA, and you can see it says Brazil's first group stage match the passion T shirt to feature Brazil training T shirt, yellow, post 2 hours before kickoff. So it's some information about when we should do our social media post and which t shirt that it involves for this particular match. If I close that again, we can see that the calendar's got the information, and we can see that it's got the information on multiple dates. So the 15th, 23rd, 24th, 18th, 25th, 19th, and the 20th are all dates in June when we have fixtures for the World Cup. So now we've got that information, we can see what Cowork can do with our calendar. So back at Cowork now, let's paste in our prompt for our new task to use our connector. So here's our prompt, and it says that we want Claude Cowork to read our Google Calendar for all World Cup 2026 match events involving England, Brazil, or Argentina. And for each match, what we want it to do is identify the relevant passion sports T shirt from Passion sports products, draft and match a social post for Instagram and X, time it for 2 hours before kickoff, and then include the correct hash tags for the territory. Then the other thing we want to do is we want to save all of these posts to Passion Sports Marketing matcheday social post dot TxD. So the project that we want to work on as usual is within the documents folder. And this time, I'm going to say, Always allow so we don't need to keep asking it, so it's going to default to the documents folder. I've left this to ask before acting just so that we can see if it doesn't need to ask us anything we can see every step. And in terms of where it's reading from, let's just quickly go to that folder. So within the passion Sports products folder, there's this World Cup 2026 folder, and then these are just our specs files, which if we open them up, we can simply have relevant details about each product as usual, and that will help us to write our social post or do whatever we want to based on real information about each of our products. So now if we go back to Cowork, we can run this, and it should be able to read all the right information from my calendar, read the information from the files in passion product sports, and then generate our social post for us. So let's get it cracking. Let's run it. So here we go. Cowork has done its thing. It loaded the skill, and it simultaneously checked for the calendar for World Cup matches and browsed the passion sports products all at the same time, which is one of the beauty of using Cowork as an agent. It multitasks. And then it got the products, read all three product files, searched my calendar in parallel. So once it got everything together, then it gave us this summary of what it's done, and it has created the match day social posts. And the summary of what's in there are seven matches, 14 posts in total. And so for England, it's got the details here confirmed for the England 2026 World Cup, T shirt. That's the skew the unique identifier for that product and the price. And it's saying that there are three matches that it's got details for at various times, Brazil two matches, and another two matches for Argentina. And it says that each match has a distinct Instagram post, which is longer bilingual favor for Brazil and Argentina and product detail, and then a punchy post on X, formerly Twitter Territory hashtags are pulled directly from the product specs, and it looks like it's generally done exactly what we want it to do. So we can now go and check that all of the information's correct as per the calendar and that it's posted it to the correct place. So let's go and have a look here. So on the left here, we have our calendar. You can see there's matches on various days, such as the one we looked at earlier on the 15th, 3:00 P.M. With England. And on the right here, you can see we can go into passion sports and then marketing it's the match day social post we want, so if we open that up, you can see, for example, in the social post, which it says it's generated, and it covers these matches here in these formats and time to 2 hours before kickoff just like we ask. And you can see as an example here, if we pull up this England calendar item, you can see it's got the details here that kickoff is at 3:00 P.M. Just like it says here, British summer time. It's got the location to say to schedule the post an hour before, and it says what the featured product is. Then underneath, there are a couple of different posts that we can see here. The one for Instagram says, It starts today, and it's got the English flag and a little football eimoji. England step onto the biggest stage in football in just 2 hours, and we want you wearing your colors when they do, England, 2026 World Cup T shirt is the only way to watch this one white base, nav and red trim, 100% organic cotton made for moments like this. Grab yours before kickoff and it tells you where to go. It's complete with the hash tags here. And then in the Twitter One, it's got a very similar message, a little bit more cut down here and with the same hash tags. So that's done exactly what we've asked it to do. And bear in mind, that's without touching anything. We've just got items in our calendar, and we've got a spec file saying what t shirts we have, and coworkers use that to just generate these social posts for the right dates at the right time. Now we've got that done, we're back at Cowork. Let's run another task, and this time using our email. So here's our prompt. It says, Check my Gmail for any customer emails received in the last seven days. Look for complaints, refund requests, sizing questions, shipping problems, positive feedback about World Cup teas, summarize the key themes into weekly digest and send the digest as an email to Team at passionsports dot code at UK with a subject, customer feedback digest, week ending, and then the date, and then wait for my approval before sending. So we basically want to gather all of customers' emails and then send a one email summary digest that we can pick up on as a team and then work out what we need to do next. Let's quickly go over to Gmail and I'll show you some of these emails. So here we are inside of Gmail. And because I've obviously had to generate some demo emails, I've actually sent these to myself, and these are some fake email addresses. But what you'll see is there are a number of emails here from different clients. For example, we've got one called Ship into Germany. Hi, do you ship to Germany. I'm a big football fan, and I'd love to order the Germany training T shirt, if you have one. Can't find shipping options on your website for Germany, Danker, Kai. So that's the first one. Very clearly it's an email to us to do with our products, and so Cowork should be able to pick up on this. The next one is a refund request Argentina T shirt, Colorn. Got the subject refund request. Argentina T color wrong. Hello, I received my Argentina fan tea today, but the color looks wrong. It's a plain, pale blue, not the sky blue and white stripe shown on the website. I'd like a refund or a replacement with the correct striped version, if you have it. Order number, and then there's the order number in regards, Fatima. And then there are four or five more, one for a Brazil T shirt, telling us this is a great product. Hi, ordered the Brazil training tea for my partner who's Brazilian. He absolutely loves it. The colors are exactly right. Not a cheap yellow like some other brands do already recommended you to FIs keep up the good work. Thank you very much, Sarah. And so we've got a few more W is my order sizing issue. Love the England tea, so we've got a mixture of good and bad ones. So what we're going to do is we are going to search through these all from Cowork, and then we're going to collect our emails, let's head back to Cowork and do that. So here we are back at our task, which is going to check Gmail for these customer emails, which have been received in the last seven days because I recently sent them, and then it's going to look for all different types of emails, summarize the key themes, and then send a digest of that. But it's going to wait for approval before sending. So let's do that now. So on the right here you can see the plan that Cowork getting through. It's read the calendar events. It's read Passion Sports product specs. I wrote the match day social post, and then we went on to search Gmail for customer emails across five categories, and now it's going on to draft a digest email. Now, even though these were, in theory from another task, I did them all in the same message box, and that's why it's include them all together. So now Claude Cowork has created the emails or it is creating the emails. And as with all things, Claude always checks because we've told it to ask before acting, and in fact, for some tasks, it will ask anyway, due to the security settings. But it says Claude wants to create a new draft email in the authenticated user emails account. I'm going to allow that because I want it to create the draft. And instead, I'm actually going to say, allow for all tasks because it's only a draft anyway. So Cowork has completed that task. You can see it searched across all five categories that it found. It could see six genuine customer emails that were forwarded in, which is correct. It pulled in the full content, then it compiled the digest and created the graft, the draft. And then the digest was drafted, and it says it's now sitting in my Gmail draft, not sent, and it's got a summary of all of the emails here. So complaint or refund, sizing or exchange, chipping inquiry, a couple of positive ones. So it seems like it's definitely found everything. Let's go across and look at the email draft that's created. I'm going to go down to my draft folder in Gmail. And then we can see here there is a customer feedback digest here from the weekending May 12. So if we have a look at that and then zoom into that a little bit, here is the digest, and it says, Hi team. Here's a summary of customer emails received this week across our five track categories, six genuine customer contracts identified. Forwarded via hello at sports passionsports dot cod at UK. What I like about this is it's categorized them into complaints slash product issues, refund requests, sizing questions and exchanges, shipping problems, queries, and positive feedback. It's stated the number of contacts for each who has actually sent in each category of email, and it's stated what action is required for each and no action for the positive feedback. And then underneath, it summarized each of the emails. So, for example, there's a complaint here from Fatma with the order number. And it summarized that she received her Argentina Fanti in a plain pale blue rather than the sky blue. It says what the action is, which is to investigate an impossible fulfillment error or incorrect stock ship and then confirm the correct stripe color weight is in stock and respond with a replacement. And then on the other hand, down here, it summarized that positive feedback. There are two Marcus and Sarah, as we read before, Marcus likes his England Wild Cup Tea and Sarah likes the Brazil training tea. It's also summarized the key themes in this week, which is the England Tea sizing running small Argentina fulfillment issue, international untapped product demand as well and strong quality perception. So it's a mixture of a few negative and some really positive signs as well. So all in all, this is a great email feedback digest that we can send to our team that tells them exactly what they need to do, where the priorities are and where the good things are as well that we need to keep doing. Now, the great thing about this email is if we go back to CW, if we look at what we've actually put in as instructions as a prompt for this task, nowhere did we say exactly how we want the email to look, but it's given us a really clear professional email that we can send to our team, and that's one of the positives about Cowork. It does what it does really well and professional. So the thing to remember about the Gmail connector is that by default, Cowork will create a draft and then show it to you before sending. This is because of the needs approval setting, and it's the right default for any outbound email because we don't just want emails going out unchecked. We are not sure if AI could make some mistake that we will regret. And you can change individual connector permissions in settings connectors, Gmail, and then to permissions. So we're back at Cowork. Let's use another one of our connectors. This time we're going to use the Google Drive connector that we set up earlier. So let's paste in our prompt. So here's our prompt. It says to read the file Passion Sports Inventory 2026 from our Google Drive in the folder Passion Sports, identify any products where stock is below 20 units and save a stock alert in passion sports reports restock alert. So let's have a quick look at our Google Drive and then look for this passion sports inventory 2026 XLS. Here we are Google Drive. I'm going to go into the Passion Sports folder. And here you can see the Passion sports inventory file that we've got on our Google Drive. So if we click into that, we can see that this is the inventory file and we can zoom in a little bit. And in the inventory, it shows the stock levels, and it flags any product below 20 units. So we've got the skew, the product name, the territory it's in, the number of units in stock, at which point we should do a reorder, which supplier, and what the status is. But you can see here that these products, which is the Morocco T shirt, South Korea, Italy and Germany T shirts, are all below and the number of units in stock are 18, 14, 11, and actually zero, respectively. They flag whether they're low in stock or out of stock. So this is the file that we want to read, but we want to do it using AI so we can always check this file from a file. We don't need to go into Google Drive and open it. So let's head back to Cowork and do that. So as I said before, it's just going to read that file and then identify products where stock is below 20 units. So let's run it. So Claude starts working as, again, permission to search files in Google Drive. I'm going to always allow because it's a read only task. So Claude is completed, and it's confirmed all data is locked in for the four products flag, and it's writing the alert. So we can even see here before we look at the file that there are four products below 20 units, so it's got that correct, and it's saying at what levels they are one that's out of stock and three that are low in stock, along with the details from the file. And then it said a couple of things worth flagging beyond the numbers. Germany is the most urgent by some distance, which it says here. It's at zero units. And what it's done because we've done all of these in the same chat, it said that from the Gmail guide just earlier this week, you already have a customer Kaimller actively trying to buy it. World Cup starting in 30 days, this needs a Kit Pro order today. I also spotted a file called Germany Spec draft Do Not Use worth checking whether there's a revised Spec pending before the order goes in. So this is some intelligent stuff going on with AI now. It's put together the fact that we've asked for a few things in the same task. I knows what the requests are from the customers, so that if we're actually flagging up things that we need to restock, it's telling us how urgent they are. That's really great. And as usual, it's put a link to the file that's generated here, but let's go and make sure that it's put it in the right place. So here's our restock alerts. If we open that up and then zoom in, we can see essentially all the same information here. It shows that for the German training T shirt, there are zero units. It's out of stock, and we've got our report also saying that there is customer demand for this. So that's worked really well, and this is going to help us to stay up to date with what we need to restock and be ahead of the game. So in summary, connectors can be set up using settings, connectors, connect, and then you authorize once and then can use in any task. In terms of what connectors or apps are available, you've got everything from Gmail, Google Drive, Google Calendar, Slack Notion, Microsoft 365, 38 plus more at this time, and growing rapidly. You can use Gmail to automatically create and draft your email and then send the email, but the default is that it needs approval. So you have to review before anything goes out, and I'd recommend just creating it in your drafts and then you can send it yourself if need to unless you're really confident. And then the connectors work inside schedule task too, so you can automate pulls from different apps and just generate these things when you're not even at your desk, which is really useful. And so now it's your turn. What I want you to do is connect your first app. I want you to connect Claude to one app that you use every day. It could be something like Gmail, Google Calendar. They're good starting points, but whatever you would find useful, you can connect. And then ask Cowork to read something from that tool and produce a useful output from it. Now notice what it's like to have Cowork acting across your actual tools, not just local files. Then you're really going to see how useful this is because these are things you would usually do every day. Now you can give it to your AI agent, cow to do all of that for you. So have fun doing that, and I will see you in the next lesson. 17. How To Use Skills To Give Claude Domain Knowledge And Best Practices: In this lesson, we're going to cover how to use skills to give Claude domain knowledge and best practices. Skills are reusable instruction bundles. You create yourself, and a skill is a short markdown file that defines exactly how cow should handle a specific type of task. You can build it once, you can upload it, and then every time that task comes up, cowork loads and follows your skill automatically. So let's get into it. So what we're covering today is called skills. Why we use it, we use it when we want to create reusable task specific instruction bundles that cowork loads automatically. And what it solves rewriting the same instructions every time and getting inconsistent outputs on tasks. So for example, if at whether you're a project manager, product owner, business analyst, and admin, and assistant, there are some things that you do probably every day or multiple times a day the exact same way. So it could be things like getting a report written. It could be creating a chart. It could be creating a product requirements document, and you do those things the same way every time, especially if you're doing things like marketing and working in a company that we are like a T shirt brand. So what skills do is they provide a way for you to create reusable bundle of instructions so that you can just run the skill instead of having to do the work the same way every single time. So let's have a look at a skill MD file and all will become clear. It is just a simple text file here. We've called it SkilMD at the moment, and MD is a markdown file, but as you can see, this text file, it basically has some separators that separate the different sections and hash tags as well that separate different titles. And essentially, this is the name of the SkillMD file to give it some context. That's the passion social post file. And there's a description here, which is to use this skill when creating social media pods for passion sports Wld cup 2026 product launches across Instagram X and Tik Tok. So this is the reusable skill or the reusable set of instructions for doing that. And as you can see, it's got various titles and then some numbered sections with the instructions and sometimes hyphenated sections with the instructions, and then descriptions under each of the headings to give it some meaning so that Claude knows what to do with this file. So let's go over the different sections and how this file is structured, and then we'll go on and use this file. So as we've already said, there is a separator in terms of these three hyphens at the beginning at the end of the first block, and we call this the front matter and the front matter sits between these markers, as I've said, at the very top and cow reads that to identify and activate the skill. That's what contains the name and the first description here. The names are unique identifier in lower case with hyphens, and it's how you reference the skill explicitly within cowork and even clawed chat. Then the description is the trigger. You write it as use this skill when and then that tells Claude when to use it. The more specific you are, the more reliably Cork activates it because that description tells it what the skill is actually going to be used for. So here you can see that the name is Passion Social post, and the description is used this skill when creating social media post for pasion Sports World Cup 2026 product launches across Instagram X and TikTop. So it's very clear the name of this, how to invoke it and what it's used for. Next, we talk about the mark down body, and that's everything below the front matter. And it consists of various sections, and then the hashtags tell you what level those sections are at. So at the top level, at the highest level, are the double hash tags, and all of these are at the same level, and then they are various levels within it. So the first one is the structure, and the structure is a numbered list of what to produce and in what order. It's the most important section because it's actually the recipe for what we're actually including in that skill. So if we look at our text file, the structure here says, first of all, in our social post skill. So when we're creating social posts, the first thing we want the structure for it is a hook, which is one punchy line, which includes the emotion, a reference to the match as in the football match that we're talking about, or a bold claim with a max of ten words, followed by a product call out somewhere in there, which is the name of the product, and then one key visual details such as the color, the design of the graphics are the white cotton England T would be one of them. It contains the name, the England T and then a visual detail like the fact that it's white and cotton. And then the quarter action, which says, available now passionsports dot cod at UK or Link in bio. So it will pick one of these quarter actions so that the user actually knows that they can buy the product and where they can go to buy it. And then that's followed by hashtags, Hashtag Wok up 2026, hashtag Passion Sports and hashtag the team nickname, whatever the nickname is for that team. So that gives it the structure. So this, as I say, is a numbered list of what to produce and what order. So now, Co Op knows exactly what to produce for our posts. The next section is rules in the markdown body, and the rules are the hard constraints. The word limits, things to always include, things to never do, rules override Claude's default. So these are like the dos and don'ts. So if we look at our markdown file, the rules here say for Instagram, there's a max of 150 words in a post. There must be a max of 150 words. For X, slash Twitter, a max of 280 characters. That's really strict and count every character. It's important to say that because sometimes Claude doesn't count them unless we tell it to. For TikTok, we want it punchy, emoji friendly and use trending hooks, and always reference the team or tournament. Never use the words amazing, perfect, or incredible. Sometimes some of these things are cliche and we don't want to include them. And then the tone should be fanf, so it should work for the fan and target what fans want to hear and someone who knows the sport and someone who's not corporate. So these are the dos and don'ts really of creating our post. The tone sets the voice and the style of whatever we're creating. So it's usually a one sentence of example language that often works better than adjectives, even though we could put adjectives here. So looking at our markdown file, the tone here it says, real fan language, short sentences, you're one of them, not a brand talking at them. So that means when Claude Cowork is writing the post as our marketing assistant, it needs to think as if it is one of the fans, not a brand talking to them because fans don't really like being sold to. So that's why it says real fan language. We could have given them an example here, but in this case, we've described exactly what it needs to do, what Claude Cowork needs to do. And next, we have the context. So the context is background about your business or your audience, and it replaces explaining that every single time. So the context is everything it needs to know to understand about the environment that it's working in or the business that it's going to be working for, the audience that it's going to be delivering the message to. So as we can see in the markdown file here, we've said, passion sports makes 100% organic cotton World Cup fan t shirts. The audience is football fans aged 18 to 45 who care about the game and want to wear something that feels authentic not mass produced. So that's enough context so that when it's right in the social media post, it knows exactly who the audience is and who it's talking to. And then optionally, we can include some examples. They're definitely optional. You don't have to include them, but they are powerful. So often one good example, output sets the bar for quality, and then Claude can just copy that and keep things at that level of quality or say things in the way that you want it to based on the example. In this case, we haven't included an example, but we felt like what it has already, the structure rules, tone, and the context will be enough to get us to where we want to be, and we can always update it if we need to. So now we've got our skill MD file. The next thing to do is to make sure we upload that into Claude desktop app so that CW will know exactly what we're talking about when we reference this skill MD file. So let's go to Claude Cowork and do that. So we're back at Claude Cowork and the way we upload a SkillMDFle is first of all we go to customize and then click on skills. And then within here, we can actually add the skills that we want to so you can see here's one that we added earlier. These are under personal skills. There are also built in skills that you can see here. But if we click the plus sign here, we can simply say create skill and then upload a skill. Then what we need to do is either drag and drop the MD file in here or click here and upload it. I'm going to click here. And then within Passion Sports, there's this folder called Skill file. So I'm just going to open it up here, and this is our skill MDFle you'll see us uploading. And you can see it's appeared here under SkillSPassion Social Post. If we click on it, it's got the structure here. Let's just put it in a bit of a cleaner structure here. You can see it was added by me, and it's got the full description here. You can also see it's been enabled so that we can run it. So now let's go back and run this skill. So we'll go back and we've got a new task here. So pasting in our prompt, this is how we run the skill. So it says using the Passion Social Post skill, so it's got the name, the exact name of the skill. Create social Post preview cards for our England 2026 World Cup T shirt. The T shirt details are in Passion sports products, England, 2026 T specs, dot TXT. So these are spec files for each and every T shirt, and produce cards for Instagram, X, and Tik Tok. So here in our file system, you can see, these are the specs if we open that up, and then we zoom in. You can see that this is the specification for the T shirt. It's just the England spec T shirt England Tshirt spec with all the usual details in here that it's going to use to create our social post. So let's run our prompt using this skill and see what happens. So now Cowork has created our post in cards, just like we asked it to. The Instagram post is here. It's put a placeholder in for the product image, but the Instagram post says, This summer means everything, England, 2026, World Cup Tea, white base, with Navy and red trim, 100% organic cotton. And again, certified by the organic standards organization. Where's something that feels real. 24 99 shop now at passionsports dot cod at UK, and it's got our hash tags. It's confirmed that it's within 100 and fm because it's 90 words, and we can see how it would look. And then we've got the X post, X or Twitter. This summer means everything. So emojis, England, 2026 T, white with Nav and red trim, organic cotton, 24 99, shop now and our Emojis is under the character limit. And then finally, our Tiktop POV, you found a T for this summer, England, 2026, World Cup Tea just dropped, white, navy red, organic cotton, real fan energy, not a souvenir, 24 99 Link in bio, some hash tags. And again, 40 words. It says it's caption ready. So it's told us that it's created them. It's given us a little bit of a description of what it did for each of these different postypes and why it did them. It's saying that all three avoid band words and follow the fan first tone from the skill guidelines. So it's acknowledging that it's actually following the skill guidelines. Now, the great thing about this is not only have we run it for this case, but we can run it for any case. So, in other words, we can re run this exact skill. And the way we do that is simply by putting in a forward slash and then the name of the skill that we want to run. So in this case, it's the Passion social post. Now, we could type in that exact word, and it would work. But because it brings up a list, I like to just put in a forward slash, and then we can choose the one we want. And so now we can just create a social post simply by giving it a command and it will create them using the exact same format here for all social media accounts across different platforms. So let's do that. So here I've said, create me a passion social post for the Cup final of England versus USA USA. I've not given it any other information. And if we run that, we can see that CWC has created our cards for each of the platforms just the same way as before. It's got the same look and feel, the same placeholders, and the same style of post because we've used the skill to do it. So now we've got a set of instructions we don't need to type in every time. Clau CoW knows exactly what to do. For example, for Instagram, says England, USA World Cup final. This is the night where the white Navy red, like you mean it because this time we do England, 2026 World Cup T shirt, 100% organic cotton, and then it goes on to show the pricing and all the other information, and it's done everything else in the same style it did before. So this is really useful. Anytime now we need to do a social media post, all we need to do is passion social post, and then a little bit of instructions, and we know we can do the same thing. And this is reusable, but also we can create any skill we want for any purpose, any reusable purpose that we want. So in summary, you can write a skill dot md file in any text editor with the name, description, structure, rules, and tone. You can even get Claude to do it for you, just tell Claude to create you the skillMDFle and what it's for and even what you want in it to save your time. You upload your SkillMDFle via Cork sidebar, customize, and then skills. And then Cork applies it automatically when a matching task is detected. It's even smart enough often to figure out which skill it needs if you don't call it by name, but you can call it by name for accuracy. And you can build one skill per repeated task type, for example, social posts, reports, email drafts, et cetera. Skills are your own custom instructions, essentially, and we'll go on to plug ins in another lesson, which are pre built packages. But with this alone, with skills alone, you have a really easy way to run the exact same instructions time and time again and create output that is going to be really helpful to you and your business, save you time, and help you to be more productive and make more money. So now it's your turn. What I want you to do is write your first skill dot MD file. Identify the most repeated writing task in your own work, write a basic skill dot md file with the name, description, the structure, and the rules and all the things that I've said. Upload it the way I've shown you and run one task using it. And then once you've done that, notice how the output differs from a generic brief. You've got now a consistent way of running the exact same thing every time, and you can change it yourself simply by adding a few instructions. So there you go. Have fun doing that, and I'll see you in the next lesson. 18. How To Extend Cowork Using Plugins From The Marketplace — And Build Your Own: Lesson, we're going to talk about how to extend cow using plug ins from the marketplace and build your own. Plug ins bundle skills, connectors, slash commands, and sub agents into a single package. What you do is you install a plug in and cowork behaves like a specialist for a particular role. For example, a marketing plugin knows campaign briefs, a finance plug in, knows invoices, et cetera. In this lesson, we're going to cover how to install plug ins, use them, and customize them, and build your own from scratch. So let's get into it. So what we're covering today is plug ins. Why we use them is to install pre built domain expertise or build your own custom plug in. The problem that they solve is building all your own skills from scratch when there's already a plug in that has the best practice workflows built in. So often what we do is we will build skills for things as we go, and we may not realize that all of those skills exist bundled up along with all the things we need to make them work. And so we don't want to reinvent the wheel and plugins help us to get whole set of skills and everything we need to do a particular job all bundled together. So let's delve into what plugins contain. First of all, plug ins contain skills. So skills are the domain knowledge for the particular domain that you're working in. So they consist of markdown files that give clawed procedural knowledge, and they load automatically when a relevant task is detected. So when you tell clawed cow to generate or post a social media post, then it knows based on what you've asked for which skill to use, it uses the natural language to detect that, or you can call the skill directly. Then we have slash commands. Slash commands are user triggered actions. So when we type forward slash into CW, it will come up with a list of all these user triggered actions, which are usually a list of actions that we could take in our current situation, and they go into the cowork input or the chat to trigger a defined workflow all in one keystroke. So it could be a particular skill that we want to call. And when we type in forward slash, it gives us a list of all those skills, and we just pick one. Then we have connect so connectors allow us to integrate tools or apps into cow, into Claude. So, for example, we have connectors for Gmail, Google Calendar, Slack, could be Jira, if you're a project manager, and all of those are connected via connectors. And now a plugin can bundle all the connectors it needs so that we install it once and the integrations come with it. So it bundles all the things I'm mentioning, including the connectors, and that makes it really helpful. Then we have sub agents. Sub agents are nothing more than specialized agents. And these are agents trained for a specific domain. So, for example, it could be project management, it could be marketing, it could be accountancy. And these specialized agents take a task end to end with deep expertise built in. So it may be that we want to apply a few different skills using a slash command, and then we want to connect to different apps using our connectors, and then the subagent knows exactly how to take our task end to end using everything that's packaged up in the plug in. So these four things combine inside of our plug in. So let's have a look at how a plugin is bundled together and all will make sense. Buttons now here. Click that. And then here, what you're going to find is all of the different plug ins supplied by Anthropic and its partners. And we can search for specific ones we want. But if you scroll down, you can see we've got plugins to do with productivity, design, marketing, engineering, and you can keep scrolling down and find what you're interested in, or you can search for specific things up here. But in this case, we're interested in marketing, so we can click here and see what the marketing plugin is all about. And it says that it allows us to create content, plan campaigns, and analyze performance across marketing channels. Maintain brand voice consistency, track competitors, and report on what's working. So you can ask different things like draft the blog post with SEO optimization, plan a multi channel campaign, review content against our brand voice, audit our SEO and find content gaps. And there are a bunch of skills with it, which is things like brand review, campaign plan, competitive brief, content creation, draft content. And you can see they're all accessible with the Ford slash, as we mentioned before. If you hover over them, they give you a description of what they do. So, for example, if I go over the email sequence one, it says, Design and draft multi email sequences four copy timing branching logic, and it gives us a background there. Whereas if I go over content creation, it says draft marketing content across channels, blog posts, social media, email newsletters, et cetera. So this looks like exactly what we want. So I can go here and click Install. Another way I could have done it was to add this plus button. And so now that's installing. You can see it says marketing is installed and ready to use, and that's within customize. So if we now go back, we can see that we've got the marketing plugging here if we were to click on that. We can see it's currently disabled, but we can enable it by toggling this switch now our plugin is completely enabled, and we're able to make use of all the skills and the connectors that are part of it. So if we now head back into cow, we can make use of the plug in, so let's go. So now we can put our marketing plug into the test. Here's the prompt that we created earlier. And before we even run this, it's important to know that we also have a file in the file system called the golden boot dot toxT. And if we flick to that quickly, we'll see that inside this text file, basically, this is a product spec sheet, so it's got the details like the product name, the SKU number, which territory it's in the price, the colors, all of the usual stuff that we've had for our other T shirts. But this is a special T shirt called the Golden Boot, 2026 English T shirt, and it's got the Golden Boot Award graphic on it. So what we want to do is we want a campaign based around this. So let's go back to COC. And so what we're saying is using the marketing plug in, create a product launch brief for our Golden Boot 2026 England T shirt. The source file should be the golden boot spec that we just looked at. And what we want to do is based on one textile output three social media caption options for Instagram, X, and TikTok, one email subject line and preview text and one hashtag set for the World Cup campaign. So essentially, what we're doing is we're using this plug in to output a campaign. We haven't told it anything except to use this spec file to do it. And then when we finish, we're going to save that to our file system in marketing a Golden Boot launch brief. We've set it up so that it's looking within documents for this passion sports folder, and we're good to go. So let's do it. So we can see that CoWork has generated the Golden Boot launch brief. It's done it in a doc file. And here it is. And we'll just read it here because we know that it consistently generates things in the right place. But what it said as a summary is that it's got social captions for Instagram, X and Twitter and TikTot. It's got the email subject line here, and it's got the preview of the text that would go in the email here. And it's created the hash set, as we've said, which is nine tags map by purpose, tournament, product team brand content type, and ethos. So these are hash tags that we can use inside of our posts. It also says that the doc includes the full product spec table. And launch reminders checklist for the team. So here is the document on the right hand side. This time, I didn't open it on a file system, but it's the same thing. So it says Passion Sports World Cup 2026, Golden Boot, 2026, England T shirt, product launch brief. And in here it gives us an overview with all of the details in the spec file. It's got the campaign themes, top scorer of race commentary, fan excitement, and tournament nostalgia the campaign themes. Got the social media captions for Instagram, Twitter, and TikTok, as mentioned. So here it says, Someone's winning the Golden Boot this summer. Put your money on England. Golden Boot, 2026 England T shirt, white base, gold trim, bold golden boot graphic, 100% organic cotton, and it's certified organic, fan first, no corporate branding. And then we've got similar posts down here for Twitter and TikTok. Then the email subject line is the Golden Boot Race starts 11th of June, 200 T shirts. Then it's gone. Preview text is here. Then it's got all of the hashtags that it's suggested that we use for this marketing, such as World Cup, six hashtag World Cup, 2026, hashtag golden boot, things like that hashtag three ions, which is what we mention when we talk about the English football team. Then at the bottom, it's got some lunch reminders to never use the words amazing, perfect or incredible in any copy because it seems a bit too cliche. No official FA or FIFA branding because those are usually copywritten and we don't want to get in trouble. Always call out the 200 unit limit because when we're promoting our T shirt, we want them to know there's a limited amount available and things like that all within the launch reminders. So this is our campaign brief at launch brief, which we can now give to anyone working in our team that guides them fully on what we're promoting, how we should post, and the dos and don'ts. A generated using our plug in. We didn't have to tell it all the rules, so it's using the expertise that's already in the marketplace. Now another important thing is that, as mentioned, within each plug in, we can actually access various skills within that plug in, and we can do that using a Ford slash. So if I type in a Ford slash now, you can see all the various skills, some of which are built in, and some of which we created ourselves. But if we scroll down, we can see we've got everything under marketing, which comes from the marketing plug in. So as explained earlier, we've got the brand review, and it gives a description as we hover over pain plan, competitive brief, draft content, email sequence, performance report, and SEO audit. So we don't actually need to run the whole plug in by saying using the marketing plugin. What we can do is access just the bits that we need. So if we want to do we want to create an email sequence, we can just access that bit of it. If we want to do a brand review, we can just access that bit of it. So that's important to know as well. 19. How To Use Dispatch To Assign Desktop Tasks From Your Phone: In this lesson, we're going to talk about how to use dispatch to assign desktop task from your phone. Dispatch lets you assign task to cow from your phone. It is running clawed. You just send a task from the Cloud mobile app, and then cow picks it up and executes on your machine. You come back to finish work. And this is actually one of my absolute favorite features about Claude, because it allows us to use clawed cow as a real agent. You can imagine, if you were dealing with an assistant or working with someone in your team, you wouldn't have to open up an app on your desktop to get something done. You could just call them. You could just speak to them, maybe via WhatsApp. Dispatch is your ability to use your phone, to speak to an AI agent and get the job done fast. So let's get into it. So what we're covering today is dispatch how we use it is to send task to your desktop cow, from your phone, and it works across both devices on the same account. And what it solves is having to be at your desk to trigger cow, and it also solves the issue of task waiting until you return to your desktop. So first of all, let's go over to our desktop file system and see what work we want to get done in our company, and then we're going to work out how we can use dispatch to help us do that via the phone without even being at our desk. Let's go. So today, we want to do two things, and we're going to use dispatch to help us to do it. The first thing is we get orders every day for new T shirts, and the daily orders go in here. So if we open that up and then we zoom in, we can see that this is just our usual orders in a spreadsheet form. And what we want to do is we want to know from wherever we are, whether we're driving somewhere and whether we're out and about, we might have some decisions to make. And we want to know how many orders we've got coming in at that particular time. So that's the first thing. The other thing, if we go back to the file system and we go in a suppliers folder, we get invoices from suppliers every week. So if we open these up, let's say we're in a meeting and we are in another building. We just want someone who's a point of contact that can quickly tell us the cost of our current suppliers. This document would actually tell us, but it's way back on our desktop. It was being worked on by one of our assistants, so we don't have it to hand in our meeting. Here, we can see if we are at our desktop that the subtotal here is 1,120 plus VAT, so it totals one, three, 44. But there's no way we would know this without looking at our desktop at that time. So the question is, how can we reach our desktop without having to look in any files and just get this back in a simple form as a sentence over the phone? So this is where dispatch comes in. We can also ask it to generate anything we want from far if we need some kind of a presentation or some kind of a file generated from this. That can also be done with the files on our desktop, and we can do all of that over the phone. So let's get into it, and I'll show you exactly how we can. So here we are back at C Work on our desktop. And there are actually two things, two pieces of this puzzle. We need clawed on our desktop. We need the clawed desktop app on our desktop, as we have now, and we need that set up for dispatch, and we also need our mobile desktop app setup as well. So let's do the first part of it first. You can do it in any order. They must both be signed into the same account. So I'm going to go over here and I'm going to go to dispatch. Now, when you start, it may have a get started button. I haven't got that here because I got started earlier. And bear in mind this is also a Beta feature. So if you look over here, you'll see it says Beta, which means that at the point of getting feedback from the general public, you and I, so things may change here. But at present, you click on dispatch, you may need to click Get Started button. And here are all the different kinds of access and settings that you will need to set for dispatch to work the way you want it to. The first thing to make sure is make sure that you give Claude access to your files, and it already has that. We've been accessing files already. The second thing is that you can only use dispatch as you can only use cowork when the computer is awake. So click, keep the computer awake, and then you can always use dispatch, and it prevents your computer sleeping while dispatch is running. So from a security point of view, just bear in mind that's the case. Next thing is, it says, Claude is ready to use Chrome, and that is if we want to be able to trigger some kind of use of our Chrome browser, I've ticked that because we used it before. You can also let Claude use your computer. I've got that off for now, because what we're about to ask it to do doesn't need the computer. It may need the browser, so we'll switch that on if we need to. And then all connectors are on, which means that all of the connectors we've attached, things like Gmail, Google Calendar, Google Drive, we've connected those as apps that it can use in the background without even using the browser, and they are on at the moment. So I'm happy with that. So once you've done that click Finish setup, and it will confirm and say that Claude would like to access the files on your desktop folder, I'm going to say, all so then you'll be presented with this screen once you've done the setup. On the left, you can see some of the sayings you've already set here. It allows you to add mobile notifications and you can enable things if you want to enable them. But I'm happy with what we've got at the moment. It reminds you that you can now dispatch to Claude and check in from anywhere a task, a code session in one continuous thread. It reminds you that you can work with Claude right in your computer. It can work with your files, browsing Chrome and use all your connectors, and we're about to set up the mobile side of it. And then here, it just gives you a quick welcome. It will allow you to test what dispatch can do here, which is essentially what cow can find confirmation in downloads and check the order status of the site, find a passport scan, check visa rules for a trip, and flag what's missing. These are typical things that you may want to do when you're away from your desktop because things like your passport, you may not have uploaded anywhere for security reasons, but you want to be able to access it on your desktop. So it gives you a few examples of things you can ask. And here you can actually test what dispatch would do if you did ask a question remotely by saying something like, tell me the contents of my documents Passion Sports folder. So, as usual, it asks permissions for anything. It says, Allow Claude to work in users Paul Lashan documents. I'm going to say Allow. And after a few seconds, it's done the check, and I can see it says, found it. The folder is actually called Passion Sports, so it didn't have the space in there, but it was smart enough to know. Here's the full structure. It tells us what's at the root level, including the Zip file we created earlier, downloads, marketing, orders, so it can definitely see what's in there. If I wanted to be notified on my phone, once I've installed the app, I can turn that on and it will notify me. Even though I've run this command here, it will tell me on my app once this has finished running, and then give me this listing. We've tested that if we do contact dispatch, it will have access to our file system. I'm happy with that. Now we can go on to do the next part, which is to install the app on our phone. So here I am on my phone, and I'm about to install the clawed mobile app so we can use dispatch. So what we need to do here is, first of all, just click on the App Store and then find the Cloud app. So I'm going to type that in now. And here it is. This is the clawed app by Anthropic. We want to check this by Anthropic and then just download. That will take a few seconds to download. Once that's done, we can get started. So there it is, and then I'll click Open. Now, the main thing is to make sure that we're signed in with the same account that we're using on our desktop. So I'm going to continue with Google. Going to and I'm going to continue because I am age appropriate. I'm not going to share my age range. And then I'm going to continue so that I can sign in. I'm going to choose the same account that I'm using n Clad Desktop App. It confirms that I'm signing back into Claude, so I'll continue. And now I'm signed in successfully into the app. So now the menus up here on the top left and then you can see on the left here, it says dispatch. It says, reach your desktop from your pocket, dispatch task to Claude and check in from your phone or computer all in one seamless conversation. So there are two ways to do it. You can email a desktop app link, or you can pair with your desktop. So I've picked pair with desktop, and it says, set up dispatch on desktop first. Download the latest version of the Claw desktop, which we've done. Sign in as the same email address, which I've already done. And then it says, Go to the dispatch tab on a desktop and complete setup. So let's do that now. So I'm going to go to the dispatch tab, and I can see that there's no change here. I'm going to tab away from it and then come back to it. Can see that there's actually no change on my desktop. And so now I'm going to complete setup on my mobile. So I can see immediately when I've completed setup on my mobile, I'm actually within the dispatch app now, and that's what's going to allow me in this dispatch section to send any messages to my desktop and essentially search and perform actions on my desktop. And I can do all of this directly from my mobile phone. Let's do a quick test. We know we can reach the file system. I'm going to paste in a prompt, and I've pasted a prompt that says, What I want co work to do on my desktop is run the daily order briefing for Passion Sports, read today's order Export from Passion Sports orders daily, and then save the briefing to Passion Sports reports daily as usual. So this is something that we can do now from my phone on my desktop. So let's run that. I actually ran out of extra usage there, so I'm going to have to buy some more extra usage, which I've done. And just remember that when you're using more advanced things like dispatch, you may use more credit and you may use more token, so you'd have to buy more. So I've gone onto my desktop and bought a little bit more. So now, based on that, I'm going to say continue. And it looks like Claude's doing what it has to do. You can see this little indicator there showing that it's working. So we're going to wait and see what apput we. Dispatch confirms that it's on it reading today's order Export, and it's generating debriefing. And while it's doing that, I'm going to enable get notified when Claude messages you. I'm going to turn that on. It asks me to confirm that Claude is going to send me notifications. That's fine. So I'll allow it. And that way, when Claude's working, when it's taking a while to do things, it will send me a message when it's finished or whenever I ask it to send me a message. So that allows me to go about my day and be informed by things happening on my desktop, just like an assistant would me. So if we scroll down, it looks like Claude finish. It says done. Here's the sum suffer today, 11th of June, six orders, 13 units, 314 pounds 87 in revenue. It's got the average order. It tells me that the English World Cup T shirt in white is carrying the day with three of six orders. So I've got all of these updates if I happen to be in the meeting that I could reach at any time, and it tells me that it's saved the briefing successfully. So let's go over to the desktop and have a look. So here on the desktop, I can see that it's saved the reports daily briefing. It's saved it in MD file. So if we zoom in a little bit, we can see the same figures here with a total revenue of 314 pounds, average order value or 52. 48. So we know we're looking at the same thing. So that's definitely worked, and that's been really helpful. Let's move on to the next one. So I'm back on my mobile now, and we're going to post we're going to paste in prompt the next instruction that we want Claude Cork our AI agent to do on our desktop. So now we've said we want Cowork to generate the week 18 supplier cost report from the files in Passion Sports Suppies Invoices week 18 and save it to Passion Sports reports, supplier Cost, week 18, and then message me when it's done. So let's get that running and then wait for Claude Cork to message me via dispatch. I'm actually going to close the dispatch app so that I'm aware when it messages me. There you go. At the top, we can see a notification, dispatch background conversation on it, reading the week 18. So there's that notification, and I'm expecting to get another notification when it's finished. And there is the message has popped up, Dispatch background conversation done. Your week 18 supplier cost report is saved to passion sports reports supplier costs weekateeent ELS. So dispatch has done its job, and if we go across to dispatch you can see that the summary is in here. It says done, and it gives us the same message. It says the What book has four sheets, a summary plus one tab per supplier. Here's the headline, two grand, 2.3 43k in total, and it gives it the breakdown with VAT across all of the suppliers, and those do look like our suppliers. So let's head back to the desktop and just check what it's actually built on that desktop. We are on our desktop and within reports, there's supplier cost week 18, and we can see we've got a spreadsheet that's just been generated here. I'm going to open that up. And if we look in here, you can see that this is the Passion sport supplier cost report for Week 18. It's got all of the suppliers, invoice numbers, dates, the due date, and the subtotals. It tells us what's paid and what's overdue. So if I was in a meeting, I'd quickly be able to see this information. And get a summary on my phone so that I know immediately where we're at with suppliers and costs. And I might be at the bank. I might be talking to another supplier and want to work out whether it's worth ordering with them. And just from my phone, I can get all the information without having to go through all of the spreadsheets to find it. And then, secondly, I can get a spreadsheet generated and send that across to anyone I need to. So that's really helpful. And it doesn't have to be spreadsheets. We can create dashboards. We can tell our co work assistant to create a webpage for us, anything that we want to do, or report, anything or anything that's useful for us. So that's worked excellently. One thing to remember is if we're using dispatch with computer use, I controlling our computer. Navigating a website or a desktop app, CoW can use your desktop computer on your behalf while you're away. You'll just receive a notification when it's done or when it needs your input. And remember that your desktop must be on and clawed desktop open for dispatch to work. Dispatch gives your phone remote control of everything Cork has permission to access, and then review your folder access and connect us before using Dispatch to make sure you can access everything you want to. It's available on paid personal accounts or P, MAX, and anything above that, and it's not currently available on Team or Enterprise at this time, although it can change, and this is in beta testing. So anything can change at this point. In summary, in order to use dispatch, you will both the desktop app and the mobile app, the order doesn't matter. Both apps just need to be on the latest version and signed into the same account. On the desktop, you'll go to Cow Dispatch and get started. You'll toggle, keep desktop awake. Make sure you've got file access and then finish setup. And then on mobile, you need to install the claw app, go to dispatch, and then you'll automatically connected. You may need to click a pair button, but at this time, it seems like they're updating it. So by the time you see this lesson, you may not need a QR code or any manual pairing. I should just work. And your desktop must be on and Claude desktop must be open for these tasks to execute. Currently, it's only available on P and Max plans on paid personal accounts. So now it's your turn. What I want you to do is set up dispatch for yourself. Set up dispatch on your phone and on your desktop, follow the setup steps that I gave you on the where to find dispatch, and then send one test task from your phone while your desktop is running. And then verify the task executes, and you can see the result in the dispatch thread on both devices. And that way, you will be able to work with your computer as if it was just an AI agent or an assistant in your company like any other assistant who can help you no matter where you are. I'm loving this feature, and I believe you will, too. So, have fun doing that and I will see you in the next lessons.