Claude AI Workflows Masterclass: Build Reusable Systems & Automate Your Work

1. Welcome to the Claude AI Workflows Course!: Welcome to the Cloud AI Workflows Master Class. If you already understand the basics of Cloud and want to move on from the one of prompts, this course will help you understand and better learn how to build reusable systems that you could apply to your everyday workflows. My name is HosnaKichui, and I'm a digital creator with over five years of experience using tools to better productivity, brainstorming, streamlining my repetitive task, and help me build more structured problem solving systems. This course, I'll show you how you can take Claude from a conversational assistant to a reliable and connected workspace. With projects, you get to structure long term work more effectively because you don't have to repeat your instructions or your context every time you want to talk to Claude. You only give those foundations once in the instructions, and Claude will use that over all the conversations you have within that space. Once we have that down, we're going to move on to Skills. You will learn how to create your first skill, something that's reusable and something that you don't have to repeat to Claude every time you wanted to do a certain thing. We will also test and refine the skills so that they get to work in various situations and give you those results that you need without wasting your time. And after that, we're going to move on to connectors. That's how Cloud can work with external apps and make your workflow more efficient. Finally, you're going to be combining all of these details into one complete business workflow. You will plan the process, select your tools, give Cloud instructions, and watch Cloud do its by the end of this course, you'll learn how to create cloud workflows that are reliable, reusable, and more efficient in the responses that they give. So let's go ahead and get started. 2. Course Overview: The focus of this course is the reusable systems that you can make with Claude. So this is more than just a chat. It's more than just a request. These are long term projects that you could use for your business, your productivity, and things that are going to retain the context that you provide and enable you to accomplish various tasks using the skills, schedule tasks, and many more things that we're going to take a look. In order for you to be able to take the most out of the course is to know how Claude works in general. I'm going to go over it very quickly, but I hope you guys have some general understanding of prompting and how Claude is different from all the other tools out there, because we're really going to be focusing on a little bit more advanced topics that really come in handy if you do plan on using Claude as your assistant. This is within the application. Now, you could apply all of these things through the web app as well, cloud.ai. The only difference is that for some of the features such as Connectors or the plugins, you are going to be prompted to download the app and do it here. So you could try it on the website, as well, but I highly recommend that you download the Desktop app. It's available for MacBook and Windows download it, sign into your account and get everything settled. You also need the pro or more plan for this as most of the things that I'm going to be mentioning are not available in the free plan. And also, you're going to be limiting yourself to a few messages, and it depends on how much you're taking from the models, if you attached anything. So if you really want to take the most out of this course, make sure that you are on at least a pro plan so that you're able to follow along and see what you can do with Cloud based on what you see. So just a very quick tour of the app, it's very similar to the web app if you just go to clod.ai. Nothing looks that different. We have our big prompt box here. You can attach files and images. We have chat and cow, the models here, voice mode, and you can even do voice to text if you prefer. There's a bunch of focuses that Claude has for writing, learning, coding, life stuff, which is related to productivity. And there's also some connectors that you can enable and that you could use through Cloud. So here we have Gmail. This is one of my connectors. Hence why I'm seeing it here. If you had Slack as one of the connectors or sauna, that is also going to be showing up here. On the top right, we have incognito mode. This is a way for you to chat with Cloud or work with it, but you don't want it to maintain that context that you're providing. So maybe you wanted to look into completely different from what you do. Just out of curiosity, you get to use this mode. It just puts you in this box, and it's going to not save your chat at all. So this is an option. It's great for certain things. We have the code tool here. I will not be going over code because there's so much to do here, and this is really something you can come over to later to experiment on but I do want to bring more attention towards the other things you could do with Claude if you don't necessarily need code or use code in your day to day work. So on the left panel, we have all the main features. We have the sidebar right here where you get to make a new chat. This is projects, artifacts, schedule task, and customize. So we have projects right here. We're going to get into these in the next lesson. But these are great for long term work and things that really need only get better with context. For example, your work, you're maintaining a full time job, you needed to know certain things about your job in order to help you within that job, you're going to get new projects and you will need the help of Claude. So a lot you could do here, artifacts are like mini applications that you get to build within Claude. For example, if you had Claude design you a workout routine that really hits a certain muscle, you can have it turn that routine into app that is easier for you to keep track of how many sets you did and how far you're progressing. So that way you have, easier user experience instead of just reading the text. Make your own artifacts here or you could have clot help you make it. That's where the cowork comes in and then chat is just a singular One main thing that you should know is the difference between chat and cowork. Over here, you may have noticed that there's a tab here. Just think of chat as a single task. So what's the weather like today? Find me restaurants in the area. These are singular one time tasks that Claude has to do. Maybe you have follow up questions, but the main focus of that session is one thing. But if you want Claude to do multiple things for you, so maybe look into restaurants, but at the same time, email three people, that's going to be cow. So it's going to be doing that on your behalf. But of course, depending on how many permissions you provide Claude, that's the main difference between these two. So you can see once I go on cowork, I can choose the project. I could choose the permissions, which I highly recommend that you keep on manually approved so that Claude doesn't do anything without you knowing. And it's pretty much the same thing. So you get to switch between these two. So artifacts, you could do it with cowork. So maybe research into X, then build build from that result, separate app Schedule task are reoccurring things that you get to do when you're not active on the application. This could be looking over your stack threats and listing the highlights for you every Monday or sending a newsletter to this many people every Tuesday, those sort of stuff where it's like an automated thing that you don't want to do it yourself. Customize is where you come in for skills, connectors, plugins, and anything else, really. So skills are basically templates that Claude is going to apply whenever you ask for a certain thing. So there's triggers in your chats, and that trigger calls for one of these skills. For example, looking at the list that I have now, have one called Excel format. So anytime I ask Claude for an Excel sheet, I trigger this skill. So this skill is going to tell Claude only use the color blue, only use this font, only include this word in the Excel sheet. And that's something that I can leave out of my prompt and let this skill take over. This is something we're going to get into in more detail because there's so much you could do here. It really saves you time when you have a long large project to work with. And again, you could do it with Claude with your own or upload something that already exists. Connectors are external applications that you get to connect to clot. So, Gmail is a different app, and I get to add it here to have Cloud maybe do something through those applications or use applications to enhance the outcomes. Plug ins are self explanatory, so there's so many things over here, but maybe if you're working with PDFs a lot, you can use a plug in to make the viewing a lot more easier. So design, if you're tired of the way Clot designs for you, you get to add a plug in. So it's a lot easier to see. It's easier on the eye. It's more efficient for sharing, et cetera. So there's plug ins from Enthropic and there's partners as well. So you may recognize a lot of these things, but, you know, there's so many that you could look into. We have Slack here, we have Miro, and so many things. Same thing for connectors. We have Enthropic and partners. So things that the creators of Clot put out and things that they have partnership with. And the same thing goes for skills. So that's your customized tab. And over here, this is where all your sessions are, and you can see that based on what I have, we have different icons. So if you see this chat bubble, that's a regular chat, so the one time one purpose slashon. But we also have cowork. So these are things. Like this is a project, for example, in something I named Irish literature. You can see that there's working folders, contexts, progress. So it's a lot more than just something you ask a question from, and then it gives you a outcome. Down here, we have design. You can design with Claude, use it in cow to aid in the outputs. For example, if you had Claude researched the fish market, you could have it visually output a presentation using design. So it's going to be more professional and you don't have to take that data that it research, put it in Canva or something to present it to your coworkers. Down here is your account. This is where you change your billing information, main information, import chats from other large language models, and anything else necessary. And that's pretty much it. So, Oh, and we also have a search bar up here if you wanted to look for something. But that's pretty much the application of clot just so that we're all on the same page. Going forward, I'm going to be focusing on some of these features because they really take Claude and set it apart from all the other application out there. We're going to start with probably one of the most important features in Claude Projects, then go on to Skills connectors after that, and then we're going to combine and apply all that we've learned onto a hands on project that we're going to do on screen. So by the end of this course, you will see Claude as much more than just a chat bot. You get to enhance its true powers, get some of the features to be involved in your daily work. And you're going to see it's going to be a very fun experiment because things can be implemented in everyday works, whether you're king, you're a student. You just want some help with managing or task. You want research. Anything in that realm, Claude can help you and be an assistant for you. So now that we have a good idea of the application and what this course is going to be, let's get started with projects. Mmm. 3. Structuring Long Term Projects: So projects are going to be suitable for anything long term. I highly recommend that you give Claude, whether it's a chat or a project, as much as contexts as possible. Because when you give a very generic prompt, you're going to get a very generic answer. And the more time you spend trying to refine that answer into what you need, you're just wasting credit and wasting usage. We're going to learn about how to properly set up projects and how to make the most out of it. So it's right over here in this tab, and we're going to get started by clicking on New Project. You're going to see this box pop up, and we're just going to give it a name. So let's say my new macha business, let's say this is my new venture that I want to do. You can add folders from your computer. For example, if before this conversation, I had a lot of photos that I took with my phone of different Macha places in New York. I get to upload them in here and then have Claude reference them whenever I'm trying to come up with a design. Or when I'm trying to come up with location. So anything you want brought in from your local device, this is where you do it. And this is where you get to describe the project and set goals. You don't have to be that in detail here because we get to do that afterwards. But this is just for your own sake. If you want to, you know, know exactly what your purpose was here, this is where you write it. So I'm just going to do coming up with a new way to maybe with a new side hustle that combines macha and, let's say, food trucks. You can do bullet points, leave it as a single sentence, whatever feels comfortable for you. So create the project, and you can see that we essentially have a chat bot, but inside this project. Now, you can think of this as, like, a little box that you're going to have in your room where you throw whatever is relevant to this topic in there. What you say over here will not kind of leak out to all the other chats unless you wanted to, but you basically get to train Claude to be the perfect assistant for this particular need. So, hence, you're seeing the instructions. I could have Claude speak to me in a certain tone, output in certain tones, only involve, maybe limit research to a certain parameter, anything else that is necessary. So over here, we get to edit the details whenever you change your mind or you want to add more to the description, you can do it there. You can archive this or delete it. We can also pin the project so that it's very easily accessed. Over here, if you have, like, a lot going on here, this is going to help. And then we have something called context. As I said earlier, the more detail and context you provide to Claude, the more it's able to help you. So this is where you get to add in anything you need, really. If it's a PDF about a certain recipe, you get to upload it here. You can also type in the context like this, and this is going to be the information that Cloud is going to reference before answering to you. So if I wrote certain instructions, and this is where you also put in rules that Cloud has to adhere to, and if I put in something like DX, it's first going to look at the instruction, what I told it to do, then the context, and then it's going to give you an answer versus a regular chat, we'll just look at your question, give you an answer. Could look all over the Internet. It could look based on its own database. There is no restraints, there's no parameters. There's nothing tailored to your individual need. So the best way to structure your long term projects is to really spend some time in these two little boxes. You don't want to go in straight ahead and have Claude kind of give you something big, and then you tailor it down because, as I said, that's going to waste credits. So let's say I'm trying to make a new Macho place, and my goal here is to come up with a branding, maybe call email to a bunch of vendors and try to figure out my brand with this idea. So in the instruction, we're just going to do the plus. We get to put in the steps that it needs to take, the rules that Clot needs to adhere to, any additional points that's going to be helpful here. We can do it in terms of, like, bullet points. Long paragraphs, even one paragraph is an instruction, and this goes alongside something called profile instruction. So before I get into this, let me actually show you what this is. You just click on it once. This is going to be an instruction Cloud has to take for your profiles. So profile instructions are going to be global instructions across all projects, all sessions, as long as you're signed in to this profile. So, for example, I can prompt basically ask it across every session that it needs to interview me before giving me an answer. Something that it does on its own sometimes depending on the context it needs. But I could make that mandatory in this box right here. And then there's a bunch of examples here. So anytime I ask a lot to research something, I could have it bold the latest data, put it at the start of every output. So that's a preference that I could set up right here. But again, it's going to be global across all of your chats. So we get to do what we here. So I'm just going to show you a quick example. So let's say, when I say Macha, you need to format your responses. In italic, something very easy. Save the changes. Saved. Let's go out. And I'm going to first ask a random question. What's the weather like today just to show you what it looks like on, like, a normal basis, gives me a visual. That's nice. And it's giving me a bunch of text. So let me see. Tell me what I should do outside today, just so we can get more text for you guys to see. So there we go, bullet points, regular text, great stuff. Should I get macho. So as you can see, the response is now italic because I asked it to do that in my personal profile. And I just had to trigger it via this word. You can always go back to your preferences. You just go on your name, then click on settings, and in general, you can alter this. So I'm just going to remove this real quick, save changes, and now Clade will act with me the way that it does generally. So that was an example. I'm just going to do Leave this guy. There we go. And then we can go back to our project. So back to the instruction, you get to add this on top of your profile, but if you want certain specific things for each project, then you don't need to do it in here. And it may even be cleaner if you leave this empty and set different expectations for every need. So over here, because I'm starting this new business, I want Claude to only give me certain parameters. Let's say, I already know so much about this drink, but I basically want Claude to tell me things that I don't know. So we're just going to do a bunch of, like, bullet points. So look into Trans in New York City. So I'm already setting a constraint. Let's actually turn this to rules. Bold all the numbers you find in your research. So let's just tell Claude what it is that we're trying to do. I am trying to start a new side hustle where I sell fresh macha out of my car, let's say, I need my decisions to be based on research and current trends. Oops. And I need you to assist me in making the right choices regarding vendors, location, branding and marketing. So rules looking trends in New York City only provide data from 2026. List me or avoid expensive vendors and anything else that you would need. So if you want Claude to do multiple things, you can list them out. You're going to help me with research with, I know, coming up with catchy phrases for marketing content, anything else. But I'm going to leave mine like this. The more important thing that I want you guys to spend some time on is the rules. So these are the constraints that you could set with Claude, and these are things that Claude is going to avoid when giving you an answer. That would mean less time for you to come in here and refine every output, but it's going to be an automatic thing where you don't get the response you don't need in the first place. So now, if I ask it, list me the current data. You know, I don't need to tell it what I'm trying to do, where I'm looking at, what's my budget. It's going to look directly into New York City 2026. Food vendor, I mentioned vendors in my instruction, affordability. That's something else I mentioned. So you can see how setting up really affects the way that you're using your credits and the way Claude works for you. So then it gives me all this list, and I have some data to start working with. I have the market trends. Now, I did not say what sort of data, as you can see, but it listed me the market trends, the vendors, and location. Those are things that I left in my instruction. So now it's like asking me a follow up if I want to make this into a plan. So it's really important that you do your setups correctly, and you can see that now we made like a subchat in the project. So you can see that in here, we have a recent chat with Claude. I could delete this. I could change the project, move it to something else, and then we could even remove it from the project. So one of the core things that you could do that's really going to help you get the best out of project, even if you don't know how to, you know, prompt properly or you don't know the correct format is just telling Codd exactly what it needs so that there's no room for follow up. So that's the instruction box. In the next lesson, we're going to go over context and providing it more knowledge so that it does a better job, you know, being your assistant. It knows exactly everything about your product and or your idea. And we're going to see how we could do it with things outside of text. Oh 4. Managing Knowledge with Projects: As we already established, giving more knowledge to Claude will really come in handy when you want to use it for various tasks, various scheduling items, and Claude can really be your assistant when it comes to executing certain things within your business. So we're going to continue off of this Macha business that we're going to start. And in the previous lesson, I just gave it a very straight to the point instruction. I gave it some rules, and I told it what it is that I'm trying to do. I said, you can always come out here, add some new details, maybe some important details about your product, what sets it apart from the others out there, and what you hope to achieve by the end of this year. So we're going to add more details here, but we're also going to look at the context box, and that's going to be the focus for this lesson because this is equally as important, and you can really add different types of media, so it could be either text like this, it could be a bunch of pictures, PDFs, even links. You can connect it to Github and give it more to work the first thing I'm going to do is make a context box, but I'm going to do it via text because this context item is going to be about what sets my Macha brand apart? Where do I source my Macha and stuff like that? And this way, Claude can use that information to tailor the responses. So if I'm going for a more local small business angle, it will avoid the big companies. It will avoid certain vendors that's not going to be suitable for my need. So I'm going to do one for my macha and we're going to give it some information. So my target audience, we can do a list like this, busy people walking around. So busy people walking around New York, you can say in need of a refreshing and caffeinated drink to boost their mood. My you can say recipe uses matcha maybe imported from Japan. And then paired with, let's say, organic oat milk with natural sweetening ingredients. So my brand is you can say minimal say brand identity. It's minimal. You can say fresh, local, and any other key terms that you think would fit here for your vision. And then we could say tone is warm, friendly, uplifting. So that's my tone. And let's actually give it a name. So my brand name we can say I don't know M Macha for now. We're just experimenting here. So that's my brand name. My target audience are these people. We could add, like a little space, so it's easier for you guys to read. The brand identity, tone, warm, friendly and uplifting. So we could say pricing, maybe affordable, maybe. So two to $5 per drink. And these are just content that I'm going to save into x. So I've used 1% of the capacity, and I could continue adding onto this. Maybe we could add some visuals as well to help Claude imagine the brand better. So for more visual context, I'm just going to head over to Canva and make a pretend presentation for this Macha brand. Say, this is a material that I've already shared with friends and families when I was trying to come up with, like, the brand itself. You can upload any other presentation, but we're just going to go ahead and search for Macha presentation. So we have a few things over here. What I'm going for is more of like a branding presentation. So, this is a good template that we can use to buy this person, but we're kind of trying to tell Claude about the branding material, so illustration style, topography, the language, even some information here, and other stuff so that it could use other basically the design context from here and apply it to the future presentations that I'm going to need for my brand. So we're just going to customize this template, and I'm just going to name it what we named it in Claude. So Mmcha and we're just going to leave that name out so it doesn't get confused because we didn't say who we are yet, so we got some style team stuff, email, et cetera, and there we go. So I'm just going to download this. And this is available for you guys, as well. Just download anything that can help Claude understand your vision. So we're just going to add it right here, upload from device. And I just uploaded the same presentation. So that's the strategy. And then we could add this into our instruction. So Claude has an additional material to pull in information from. So I could do another rule when asked for a presentation referred to PDF let me see what I named it. So refer to, and then you can put the name of that presentation PDF name. I didn't change the name, so I'm just gonna put something long. So Claude knows what to do for style, formatting and colors. You could do another one when forming written text, maybe, and other materials refer to the I don't know if we named this M Macho or Mmcha. I think I did Mimcha. Save instruction. Okay, so we named it my Macha. Another thing that we could add here is a sample of us speaking to someone else via email so that cloud can understand my tone of voice, the way I communicate with clients, and it could use that exact same language. So this could be another content. We can say my tone or just tone. It could be you can just write any text here. Hey, guys, hope it's a great day for you. I'm going to use things that set this text different from generic text just so we can see it in action. So a lot of, like, symbols here. I want Claude to kind of understand and study the way I speak. So try to really put your personality here into text, and that way, it could copy and kind of maintain that speech throughout emails, throughout presentation development, documents that you're going to need, brand strategy, and other stuff. So something like this, I don't know, it's so much fun, and I love mata. I'll, like, finish with best to you all, something unique. So this is a sample. I'm going to add this content, and we're going to add that to the instruction. Another bullet point, use tone when forming written content. And then we can say, and that's it. Save instructions. So we got a bunch of stuff going on here. And now I could just go into chat and have it generate a caption based on, you know, the content that I just provided. So I'm going to go back here actually, and specify the written content. So use tone content when forming, I could say social media captions, emails, and presentations. There we go. Alright, so now we're going to write me an email that I could send to my audience. So write me an email that I could send to my audience, which announces a new macho flavor. So let's see how it's going to incorporate that speech that we just gave it into a very simple email. So matching Brandt's casual tone, as you can see. Now, of course, the more tone examples we give it, the more it could help us out with. But you can see that, again, it's only looking into New York City 2026. These are instructions that we established in the previous lesson. So it's just trying to come up with macho flavors right now. I could have also said it here, but, you know, more research for us. So you can see how it exactly copied my way of writing, my tone, and basically the energy that I set in that content. The dabble exclamation marks, same thing here, there's the way that it's speaking, and some details that I talked basically gave Claude about my brand. So straight from Japan, organic oat milk, natural sweetness. These are things that I mentioned, the same terms, refreshing, caffeinated, busy days, and even the price point that I told it that it's for my brand. Of course, my sign off is the exact same. So now, because we are in chat, it just gave me the email with the subject line. But in the next lesson, we're going to see how the cow feature would work here. You can see that it gave me this, and then there's a button to open this in mail or send via Gmail. So I just have an extra step to do right now because this is via chat, but we're not going to worry about this right now. I just wanted you guys to see how a little content really helps Claude better understand your vision and execute based off of that. Going back to this, we now have two chats. I could try something with the presentation. So we can say research some flavors, and then we could apply that in the presentation. So let's do the plus here. We'll do research. Research and come up with three unique flavors that would suit my brand. So because I said my brand, it needs to go for natural sweetness. It needs to go for refreshing, and oat milk is one of the other key features. You can see right now because we said it on manual, I did say that if you have that on manual, it's going to ask you for permissions. So this way, I know that, you know, I don't need Claude to use GML for researching. I could simply say skip. But basically, if we had not put it on manual, it was going to send an email to, I don't know so that's why it's really important for that to be on manual, so it asks you and prompts you to say yes or no to a certain action. So right now, it's looking straight into New York City. If you click on this guide, it's going to show you exactly what it's doing. It created a research plan. It's looking at 55 sources and counting, and then it's going to use its own intelligence and, you know, details from my contact to form three unique flavors. So this may take a while research, it's kind of like sweeping the entire Internet, looking for a specific so don't be alarmed if this is taking a little too long, but just give it enough time because as you can see, it went from, I believe it was like 30, 40 sources, and now it's 100 something. And we can just see the time that is elapsing. And up here, it's telling you at which stage of research it is in. So we're going to let this run. I'm going to see what sort of flavors Clot comes up with. We're going to refine it based on our preference, maybe have it do more research for a certain ingredient or what is in season, and then we're going to continue applying this over presentations and see how we can incorporate a project in an external session within clot. 5. Reusing Context Across Tasks: Okay, so in the previous lesson, we had this Macha project where we gave it more context. We talked about our branding. We even uploaded a presentation that tells Claude exactly our style and how we want everything to be visually. So we ended up having to ask Claude for some three unique flavors, and we had it researched via this research tab. It finished it. It took 6 minutes and 8 seconds, and it looked over 254 sources. Once we click on this, we can see that it only looked at the places that we restricted it to. So you can see how helpful this is because if it ended up doing 6 minutes of research from random places, you would have had to do a lot of adjustment later on. So another six minute, another 6 minutes. You're just wasting your time. So research report is ready. That's this next document that it provided. I'm going to click on that and we have even the too long Didn't read section. So a quick summary, like an abstract for this. We have a nice title, key findings, details about the New York City 2026 market, and, of course, the flavor scan that we asked for and some recommendations. So it looked at everything that there, what are trendy items? What's needed? What's the market like? And then it gave us three flavors to consider. So I could look at this, say I did have this business, I could even disagree with the flavors, say that, you know, I don't want any strawberry at all, and that's something Claude can do. So recommendation, it's giving me different action plans and, of course, some things that I need to consider. Let's go ahead and talk with Claude a little bit based on this finding. I'm going to have it exclude strawberry, so exclude strawberries from the report. Say, I have a strawberry allergy and I don't want to be serving strawberry. That's something Claude can pick up immediately and rewrite the report for you. Now, the report itself, you can copy the thing. The text, you can download it as a PDF, a markdown, or you could publish it as an artifact, which is something we're going to look at later. So I'm first going to try altering the research report that Claude gave me. You can see that it's drafting the artifact right now, but I'm basically going to have Claude turn this huge chunk of text into a nice visual presentation that uses my branding. So it gave me this thing again. You see that it's still not done. So I guess it's just showing me before and after Version two latest. So once it's done, it's in Version two. We still have access to Version one, which is something I really like about Claude. You don't lose your work in these projects ever. Alright, so it's done doing the little alteration, and now we have this new document that we can look at. So let's scroll down to the flavors. We have peach, watermelon, and lavender. So no strawberries at all. And it even gave us a script, essentially, that we could turn into multiple things, such as an artifact, a direction or, like, PDF, and we even have a code section. We needed that. So let's go ahead and have Claude turn this into a nice visual presentation. Because we're within the project, it needs to adhere to the restraints that we provided. So we can say turn this report into a presentation. That I could present to stakeholders, maybe. So here I didn't really talk about tone, but you could say use the tone that provided with or completely ignore that language and go for something more professional. But right now, I just want to focus on the visual aspect, and you can see that it's reviewing documentation and matching the brand styling approach. We even have some presentation skills that it's looking at. So the skills that you see right here, these are things that you may have had available to Claude to you. These are certain instructions that Claude will use for certain things. For example, we have web Artifacts Builder. Anytime you ask for a web artifact, Claude is going to look at this guy to see what tools it needs to use, how it should look, what process it needs to take. We even have stuff for Excel, presentations and other stuff. You can build skills or you can get them from enthropic themselves. This is something we're going to focus on in a further lesson, but just in case you didn't know what this is, it's pulling information from the skill list. This is basically looking at the documentation. It's building the slides. And if I click on it, I can see exactly how it's reasoning. So it's telling us exactly what it's doing and, you know, it's going over. So first, it's planning the presentation itself. So slide ten, slide 12, it's now going to generate the script, and it's really fun seeing in real time the process that Claude is taking. So going back here, you know, while something is happening in one of these chats, you can always come back to the project, do something else. But now I wanted to kind of take that presentation that we're building at the moment. It's based on this sky and have Claude do multiple things via cow. So remember in the previous lesson, we got a pop up about Gmail. This is what I meant. So because we have it on manual, it asked me whether it could use Gmail for research, and it didn't do it on its own because I didn't have this setting on automatically approve or skip approvals. So let's see where it is right now. It's still building. But with Cork, we could connect certain things that we want Claude to use. So say I wanted to email this presentation to, like, three emails, I'm going to need Gmail. If I wanted to upload this presentation onto Google Drive, I'm going to need Google Drive. So these are the connectors that you get to add. On here. So clicking on that, we can see Google Drive. If I want Claude to use the text but make the presentation in Canva, it's going to need Canva. So that's an option that you can take, but I'm going to go over connectors later as well. For now, I just want you guys to see, like, a very simple output, such as, you know, research, text, alteration, and then the presentation. Okay, so this did take a while, but it did do a lot of quality assurance. As you can see, QA, it's everywhere. And we have our final presentation. So we have a 12 slide deck, and it's telling us all the titles. And as you can see, it's using the same color and the same even font is very similar here as what it provided what we provided a Claude. Only difference here is that it couldn't regenerate the illustrations, and that's something that you get to fix easily with one of the connectors. Say you have a Google Drive where you illustrated things yourself, you just name it according to the slide, and then Claude can pull those images out, place them on the relevant sections. If you want to use another AI generated image tool, you get to add that as a connector. You could use M journey or Firefly to generate a simple illustration here, and connectors really help in this case. But things that Claude can do is take the font, the colors, the layout style, and, of course, the information from the project itself. So you can see that my two to $5 Mmcha is right here. And then the flavors that we have is we had Claude research, and then we altered is listed right here. So I could now download it as a presentation, but thing that you could do is add this as a PDF into the project as additional content. So let's say convert this into a PDF file because it gave me Power Point only, and I can't upload PowerPoints here. So let's have it do that conversion, and then we could easily make this a space. We can add this to project. Now, you can also continue designing this in clot design that gives you more flexibility over the visual aspects. But if you just want a quick visual, you don't really care about the way it looks, you can keep it within the chat. But this option also pops up as well. So now when you go back to the project, you can see that the stakeholder PDF is here. I did a duplicate. So there it is. This is the context we provided, and this is what we made within one of the chats. You can use project across various sessions and only refer to what you did here with one live forwardslash. So this right here is our pinned project. Everything regarding our Macha business is now enclosed in this project. So right over here, I just made a new session. You just click on This guy. We're going to switch to cow because we want clot to do multiple things from one prompt. But before we submit anything, we're going to choose our Macha project. So we're going to add this project. Difference here is that now that I started a new chat, I don't need to re explain everything about my business, and I could simply refer to it by selecting the project right here. So now I could do a very simple prompt. I need help. Writing emails. You know, very straight to the point. I just say what I need. You can see that the project now went inside the context box. So what kind of emails are we writing and who is it going to? So let's say I need to write emails to three people with a research report presentation. It needs to be convincing, and I am looking to speak about funding the new product launch. We now can give little the different emails. I'm just going to do a bunch of random ones, Katampledt com, c at example.com, a at example.com. And I'm going to provide more context as to who these people are. I could do, like, a dash. This is like my manager. Let's say, this is my financial officer, and this is my marketing assistant. So three people that I need this email to go to. So we're going to tell Claude craft three emails for each address, attach the report and show me the drafts so that it doesn't attempt to send the emails on its own. So you can see it's finding files. These should be within the project. I just hope it chooses the correct presentation because we had two. It's reading the files, and we can then have it alter the emails if needed, and then just send it on our behalf. So I don't need to open Gmail. I don't need to anything, basically. And this immediately went into the project. So notice how I didn't have to go in the project. I just chose it down there for cowork and it immediately understood that this is related to this project. So that's like an auto organizing feature that it has, and it's pretty helpful. It just asked me a question for clarification. I just said, keep it open ended. It could ask you more questions, and you could even have it ask you questions within your prompt. But that really helps take your outcomes to the next level because Claude will have everything it needs. You may not know in the first place. So here's my research report. So it has the exact document that we made in the previous chat. It's keeping the tone. Here are the three emails and it's attaching it, and all of that. You email connectors can't attach. So let's actually go to our Connectors. Gmail, going to manage connectors, see why I can't attach things. So just click on Gmail and we're going to have it attach. I have to just find that real quick. I'll do this one is custom, and that's it. Okay, let's go back. Just make sure Gmail is here. If it's not there for you, just go to add Loop Browse Connectors, and GML should be right here. You click on the Plus, sign into whatever account you want Claude to access, and that is it. So let's close that. Say send the emails. To my Gmail draft. So again, not exactly sending the emails, but more so putting it into my draft. You can see that is happening with the GML connectors now. So the second email and then the third email. So here on this side, we could see that it made the draft. And if you go into your Gmail, you just sign in, go to your draft. You can actually see these being generated on your behalf. So that's an example of cow. I could have it send an email and at the same time upload my document to Google Drive. At the same time, announce the new product into my Slack and do a bunch of other things as long as you have that connector setup right here. So now that we know how projects work, we went more into detail with a fake macho business. Next lesson, we're going to look at another very important feature within Claude, and that is the Skills. So we looked at it briefly when we were doing stuff here, but I'm going to go into more detail as to how to get new skills, craft your own, use Claude, to craft your own, and what you could use them for. So let's go ahead and see what those are. 6. What Are Skills?: Let's learn more about one of Claude's most unique features, and those are called Skills. Now, skills are a unique set of instructions that you can either write yourself or download some from the community that tell Claude to react a certain way whenever it calls for it. So, for example, if your company has a unique way of summarizing documents, maybe there's, like, logos that need to be in certain places, certain words, and other things, you can generate a skill made for that and simply call that skill whenever you're working with that document. So this really saves you time because you don't have to write a giant prompt every time. It doesn't really rely on the context across sessions because the skills are saved in one separate compartment. To access skills, you just have to go to customize right here and then go to Skills. So there's some here already, but I just want to show you the different types of skill that you could write or download, just so you get a better idea about how this whole thing functions. Some of these are from Enthropic the company that made Claude, and some of them I wrote them myself. Now, the ones that I wrote are very straight to the point. I think a few sentences maybe. Enthropic is obviously going to be more in detail, and I'm going to show you a few ones later on where they're just a couple pages long. So we can analyze those together. So if you go to this one, for example, you can see that I made a skill with one sentence. No detail. I just have it do a certain thing. So you can see that my skill is just for equation formatting within linear algebra, and you can get started with one single line. Maybe a couple more if it needs it. But this turns into a code that Claude executes every time you call the linear skill. Now, let's examine something that is a little bit higher up, and that is the morning skill by anthropic. If you want to download this, it's in the brows tab. So if you go to Skills, there's a few that you can get, but let's see what this looks like. Can see that it has a context section, and we're looking at skill dot md. There are three files in this, and if you ever wanted to download skills from the Internet, this is the file that you're going to have to look for. So first, it gives us a description. That's exactly what this skill is going to execute. We have a context box. We have a seta what Cloud needs to gather, meaning that this is going to be the little interview that Claude does. But we're telling it exactly what it needs to check for before proceeding. We have some priority stuff. We have some sorting instructions. Writing, visual anchor, important things, how it can build it. So you can see that we're getting some code here. We get some verification, design, even we're putting in certain colors. You can see there's orange, white, Apple systems, fonts, and some ground rules. So this is obviously way longer and way more detailed than the one I made for linear algebra. But we're just going to see the difference if I were to use the skill Wsus if I didn't. First, let's understand what this skill is for. This is a 32nd morning glance. So draw a hand sketched single file HDML page. Ddrawn as terrain. So it's like a little design. Bottom is important thing, what needs me, what is already sorted, and other stuff. So I need to make a prompt that is similar to this, just so we get a fair comparison. So let's go ahead and just copy this, and then we'll change the wording afterwards. So let's make a new session. Chat, and I'm just going to paste it here so I could reword it. So I could say create one HDML page with a warm hand sketched style. The top half is drawn as a terrain with a few words underneath. The bottom half is important things what is needed, what is sorted, and any extra sections I later call for. So I try to just make this instruction a lot more casual, something that you could just, you know, ask Claude on a random day. So I'm asking for this, like, a specific type of an HDML page. Obviously, this is for productivity. But let's see how Claude is going to work based on this prompt. I have not called the skill yet. So just put in the prompt. It's going to architect something right now. It's actually looking for skills. Because I asked for a HDML page, so it's going to need that skill anyway. And this is a general skill, meaning that it's just how Cloud is going to approach HDML pages in general. So this is not really something customized. Like, I didn't do this right now. But if you click on it, you can see exactly how it's going to tackle with intentional visual design, new user interface or reshaping an existing one. So it's just telling it design principles, very basic things, and we just have clots branding. So the orange and the white. And some more design instruction. So I could save this skill for myself, maybe replace the colors with my brand colors or have it use a certain coding language, you know, anything unique to me. But this right here is just Claude's own general skill. That's going to take a while. I'm just going to create another chat where I'm going to call for that morning skill that we looked at. So sportsla Morning and I'm just going to hit Enter. So I don't have to say anything here because it should all be inside the skill. It's going to load it and execute that based on what it knows. Now, if you have your connectors, you can see just asked about that, those things will be integrated into this skill. So I'm going to allow this. I don't have that many connectors for Claude, but if you had, like, a slack, you had notion, those things, those information will be pulled in. So you can see that it's asking for more connectors because the skill that we looked at needs Google calendar to list, you know, important events or anything needed. But I'm just going to show you the first chat that we had, and it's currently done with the HDML. So let's click on it, and this is what we have. So you can see it randomly just decided that this is going to be for a cabin trip. I did not really call for this, but there's, like, a checklist for me, things that I already did, and it's really cool and interactive. I could add more sections here by just chatting to Claude. And if needed, I could download this file or publish it as, like, a mini app. So that's that. Let me just go ahead and connect my calendar here, and then we get to see how that's going to execute as well. All right, so I connected my calendar, and immediately, you can see the permissions that it has. Feel free to pause and look at the permissions you currently gave Claude and adjust it to your need. So let's put that away. And I'm just going to redo this. So this took a while, but you can see the process from start to finish. So it's telling us exactly what it did. And by the way, if you click on any of these, you get even more information about the steps that Claude took to execute your command. So at some point, it saw that it needed more connectors. So I did not connect my Slack and Microsoft 365 before, but you can see that it's just suggesting that these two would help if they were connected in order to build this HTML page. You could just ignore this or connect it if you want. But let's see what this page looks like. So once you click on it, we're going to get this very minimal design. And I have the date up here, a little text from the top. And this is based on one of my events in my Google calendar. So you can see it translates that into something nice. We have a little calendar breakdown, and we have something that is resolved. So this is an email account that I don't really use, but if this is your daily email, you would have maybe a lot more detail on here. But still, it's interesting because it read one of the emails that I hadn't opened, and it even saw the follow up email where it says that this issue was resolved. So for context, there was, like, a tornado warning here, and then we get I got this initial email telling me to, you know, stay inside. Afterwards, another email came telling us that, you know, it's all good now. So you can see that Claude was able to examine both emails and then show us what was resolved. Now, the cool thing about this is that because the design is written within the skill, if I redo this in another chat maybe three days from now, it's going to look the exact same. Whereas this guy is going to look different every time unless I say it in my prompt, because this is a random prompt that I just put in. I did not mention any colors, and another time that I'd say the exact same thing, Claude may use blue. I may use a random color. Or even do a whole different UI. So you can see the difference between having to just do a forward slash for one of the skills versus seeing that task that you want Clot to execute. So this is just a general understanding of what a skill is. Let's go ahead and see how we can build a skill, how we can refine it, and even see how far we can go with a skill. 7. Creating Your First Skill: So how do you build a skill? To do that, you just simply go to customize in the left side panel, go to Skills and click on AD. Now, there are some options once you click on this little bar. You can create one with Claude, meaning that Claude is going to activate its own skill creator script, and you will just tell Claude what sort of help you need and what needs to be reusable. Also write the skill instruction, so it could just be, Hey, do this, this is Contacts, and anything else. You can upload a skill with a dot md file. That's something you do if you downloaded something from the Internet. And lastly, you can record your screen. So with this one, Claude is going to look at your screen, look at an action that you're going to take, record that, and then execute it. For example, sending an email and C seeing a certain mailing list. That's something that Claude can do from now on, and I don't have to do it on my own. Let's go ahead and do one with this just because maybe some of you don't even know how much context you need to provide in order to get a good skill. Let's click on this. We're going to get text, just appear out of nowhere, and we're just going to click Send. You can see that I from Clouds action. I'm asking Claude to interview me and you can see that immediately activated the skill creator skill. Now it's going to basically consult me and see how I could help, how it could help me, and I could just tell it as much or as little as possible, and then we can just go back and forth. At this point, you can think of an action that you do constantly that Claude can help you. It writ in a certain email? Is it formatting? Is it summarizing, research, anything else that you could think of? This is the place where you get to tell Claude about it and have it generate you a skill. So for me, I'm going to build a MLA format skill. Basically, any text that I provide Claude, Claude should be able to format it in the correct and most recent MLA format using the references that I'll provide, so it can just draw information from anywhere. It should be from maybe Purdue hour and then use that information to format any text that I give it. Being consistent and delivering me, high quality content. So I'm going to communicate that in as little as possible, just so we can see how Claude is going to take that and turn it into a skill that is of use to me. So I'm going to say, help me with formatting text with Emily format. That's it. Online. I'm going to have Claude do most of the thinking. It's supposed to ask us follow up questions as well. So that's going to determine the trigger conditions. You can set triggers on your own. For example, I could have the trigger word of this format be potato irrelevant, but you get to set that trigger. So what should this skill help with? I'm going to say this guy, this guy, basically all of them. If it's not in the list, you can write it here as well. What output format should it be? A document, plain text. I think I'll do PDF. So something out of this list. Set up test cases to verify. Yes, we're going to set those up. So a little interview between me and Claude, and we can now see how it's going to build that skill for us. Alright, now it's asking us about the edition. So I'll do the ninth edition. Ser gives incomplete source info, should the skill search the b. Let's not do that. Full paper formatting, what's a typical starting point. User uploads an existing Word text draft. I think it would be this one, so something I paste into Claude, not something I write right here. So two interviews so far. Let's see what it does. So now it's writing a skilled at MD file for this, and it's actually making a reference section for the full citation templates. So this is something it did on its own. I did not say anything, and it's currently creating the file. We can see it being built. So we're getting a name description. Can't really see it here. There we go. So this is the skill so far. You can see there's a lot of information four sections so far, and we even have the code preview over here. Now, it's looking into, you know, all the templates needed for this, and now it's going to do that testing that it asked us in the beginning. Is my skill so far. I'm going to hit X. So it's going to do it's going to test itself here, actually. It's going to make some few sample cases, and it'll tell us exactly whether or not the skill was made correctly. So there's a test script, validating the format. If you click on it, you can kind of see all the stuff it's doing. It's building the test script. So page one, page two. And it's really interesting to just look at what clots process is for this. Normally, it's kind of collapsed. So it's this one is what you would normally see. But if you were curious as to what it's doing, just feel free to click on any of these guys, and it's going to open up what it's doing. So this is like a sample, work cited page. You can see that it does have that Emily formatting put out correctly. So it's saying it looks correct. Test one worked well. It's just reporting to us, and then it's going to do test two. And while that's happening, if you're not familiar with what Emily formatting is, because I keep mentioning it, it's a certain way that you format a literature text and the way you format your sources. So in writing, we have multiple type of formatting. This is one of them, and this is just one use case that you could use Claude Skills for. So there's my Test one full paper. We go. We got the little header opening. We're excited page looks good, and test two is with incomplete info. So instead of guessing, the skills flag what is missing. Let's turn this off. And then we got four tests, actually. Looks pretty good. And there's the full paper. So based on this, I could tell it, you know, you missed a spot. Maybe I want you to switch into another style completely. This is the place where you would do that in the same skill. You can even rename this into skill generator. But essentially, now, I get to tell it whether it should pack it up into a skill for me or continue working on it. So I'm going to say it looks good, and that's it. So I'm not really talking much here. I'm having Claude figure me out. Now we have our dotmd file. We have the template. We have the skill itself, and if you just go over it, there's a lot of information. There's how Claude is going to build it. The first thing it's going to do, it's going to read the instruction that it just wrote. Then it's going to go over the header, head or block, resize a document if needed, convert it to a PDF. That's something I said I need. Verify it visually, drop the final PDF and present it to us in the chat. Certain rules for the work cited. Use this version. That's what we specified earlier. There's intach citations and so many other stuff. You can choose to save this skill to your account or, you know, start over, delete it, whatever you choose to do. But if you download it here, you get to share it with other people because it's just something you can pull on your local device. But if you save skill, it's going to happen within the same application, so you don't have to leave at all. And now we have one made by me called EMLA format. So you can see the difference between this guy and this guy. They're worlds apart, and this is one way that you can start building skills. You don't know where to get started, but you have an idea of what you need this. Claude can help you do that. In the next lesson, we're going to see how we could upload a skill, and then finally how we could write a skill and like, what should be the general template. What are the sections you should have and how you can apply it to different tasks. I will also show you where you can even download a skill. Where are the communities, what you should search for, and how you can get them. 8. Improving Skills with References: So let's see how we can write some skills using the second option. When you click on this, you're going to get a pop up. You should give it a name, a description, and then finally the larger chunk of text. So we're going to write the name. So I'm going to call this Emo format to description, a format, that takes links to articles, reads, and then formats or outputs. They formatted text using Emily ninth edition, guidelines, just guidelines. Now, here is where you get to either say something with one sentence, three paragraphs, three pages, you know, really depends. But I'm just going to use my natural tone of language to write instruction. So you don't really need it to be that long if you know exactly what you want and you can express it into simple word. So it's going to do a couple of things, but I'm first going to give it some context. So you can do it in all caps. So let's say I'm a college student that needs files to be formatted into proper Emily for research papers and other literature pieces. The formatted document. Got the R formatted document needs to be of best quality to ensure good grades. You know, a very simple context. You can be as detailed or as brief as you want. But provide the context and then move on to the steps that this skill needs to take. So you can do process or steps, whichever you want. One, analyze the provided the user provided link. And then I'm going to do a column here and do some sub rules for the process. So check the validity or maybe credibility ability of the source, reject any Wikipedia pages. So that's step one that it needs to do. Step two, summarize the information and form paragraphs. Step three, create a dot Dx with the proper EmiA template or header, meter box, or excited place paragraphs into the dot docs, and then five just to make this very obvious. Place one or place this sentence at the end of the document. Like, have a great day. And then rules, just to make this even more strict, only rely on Emily ninth edition guidelines. Make sure application is up to date. Rule number two, avoid maybe reject any wikipedia.com links. Inform user of source not being credible. This is just an example. Number three, use times New Roman font with 12 points and a doubles based document. And then, lastly, end all documents with the sentence. Have a great date. So that's that. Here you can provide some links for additional context. So for the Emily ninth edition, I'm just going to paste a perdu hour link. And this is a great way for you to include any other reference links that would help Claude. Guideline out there that could help you or something you wrote previously and you wanted to use as a guide. This is where you do it. I'm just going to grab that link, and let's put this in one of the rules. Rule number five, only use Perdue OL rules for MLA formatting. This is a little bit more in detail compared to the linear algebra one, but you can just go on with this. If this is a visual document, you can only use certain colors, certain fonts. You can have it use images from one website, maybe pexels.com, and just go about building as much as you want in this instruction box. Well, let's go ahead and hit Create. And now I have Emily format too. So I have two things, one that Claude wrote, one that I did. We can go ahead and test these out and see which one basically what their differences are because we pretty much ask for the same thing. So I'm going to make a new document, and I'm going to grab a couple of links. One of them is going to be from Wikipedia. We want to see if it will reject it. And the second one could be, like, a news article that I could put in here for EMA formatting. So I'm just going to go to Wikipedia here and look for roses. First one is just the Rose page on Wikipedia. And for the second article, I found this guy. It's about a very generic topic, and I could feed it this PDF, have it reformat it in Emily format, and see what happens. So let's download this guy. One of them is a file. The other one is a link. You can find the document in the resource bag if you want to follow this along. But the rose one, you can just go to wikipedia.org and then type in for Rose. So I'm first going to try this link. Let's go back to Cloud. And I'm going to do forward slash. Let's do the EmAFmat. The second one because I said, avoid any Wikipedia and just paste this here. Let's see what happens. So it's going to load the skill that I wrote. So there we go. So the page I give is a Wikipedia page, and the rule says to reject anything that's Wikipedia. So this is good. It works, and then I could just to double check this, we can grab another link that is not Wikipedia but about roses. So I have this other source by Britannica. It's about roses again, but it's not Wikipedia. Let's see how Claude is going to react to this. No text, putting in the thing. Oops, I forgot the forward slash. Let's stop that. I'll do forward slash. EmoAFmat, paste that link. So first, it's going to look at that the link. And now it's looking at the dot doc skill because I did say, create a document. It's going to look at the EmoA requirement and then break it down into a file that I could download. Now, at the same time, you can run the other skill that clod made on its own. So let's go ahead and do that with a document so we can compare them. It's like a new file right here, upload that document we downloaded. Going back to this, we can see that check that it's not Wikipedia, and then it says looks correct. It regenerated this into EMA format. There's some, you know, templates left for us to fill in for student, instructor, other stuff. We also get the work cited page. So we went from this document. You know, in a completely different format. And with one skill, we turned it into an EMA format. So this is the skill we made ourself. Let's look at the skill that Claude made for us, basically. And it's going to ask us a few questions. So just this one. It's going to need some information from me because this is how the skill was set up. You can see how different it is when you make it yourself versus when Claude takes over. So skip I'm just putting in one file and I want the so it's going to need something for my name. Let's say, that's not K. Oops, and then two can just be the default for today. This took a little longer than the one we made, but eventually it did form the document. This is what it looks like. It's pretty good. Double space Times New Roman, everything else that we talked about with Claude. This is it. Now I get to take this document, do my own editing, or do it within the chat right here. So as you can see, making skills is not that difficult. You have the help of Claude, and you can also use the template that I showed you where you just give it some contexts, some rules, and some processes. You can change or include something additional in your skills. You saw, I even had a one line skill for linear algebra. So there isn't a rule for how long or how short your skills need to be. However, with AI, as always, the more context you provided, the more words you put in, the more detailed that outcome is going to be. Now, let's move on to our next lesson where I show you how you could download and upload skills and where you can find some really cool things. 9. Where to Download External Skills: Mm hmm in the previous lesson, we saw how we could write our own skills, and before that, we saw how Clot could help us form skills. Now we're going to see where we could download some pretty interesting stuff and how we can use skills to boost our productivity. So the first place that you could go to is GitHub, and this person does have a lot of cool skills that they have made available for download. So all you have to do is go over to github.com slash Oprah SuperPowers. There's a lab version and then just Super Powers in general. And this place not familiar with Github, you can download them from these repositories, and you can just find them here. So skills that you get to use and import into Claude. So just click on that, and we have a lot of different skills available to us. These are all made by people. So for example, we can use brainstorming. We could use writing plans, and there's some code review ones which are pretty interesting. I'm going to go with this one. Executing plans, and all you have to do is download that skill dot md. Just click on it. We can see a preview first. So you can see we have an overview of the process, different steps, and there's some rules at the bottom. So let's go ahead and download this raw file. And once it's downloaded, you can see that it's called skill dot md, which is precisely what we made in the previous lesson with Claude. Let's go ahead and make this basically import this into clot. Just go to customize Skills, add Upload. Just drag it right here. There we go. So now I have one called Executing Plans, and the author is M. That's something you can alter. And when you come in here, you can edit it further. Maybe you want to add in a few details, you know, make some other changes. You get to either edit it yourself, edit it with the help of clot or just try it immediately. And you can also uninstall it if you changed your mind. That's one example. And like I said, Enthropic has a lot of skills that you can just look at. If you click on Browse, there are some available right here. We have Brown. We have something with algorithms, canvas designs, the morning one that we did and even learn. But let's take a look at some other places where you can find skills and download from. So if you search free Claude Skills or even paid ones, there's a lot of websites out there that you can take a look at. So this website is also interesting. It's like a Skills market. And a lot of them are free. So this really depends on which category you want to use clot for. There's marketing, productivity, development, payments, and many more. We have, like, a scientific skill, asient card, and a lot that you can examine. So let's go to this one where you have Clot act as the chief marketing officer. You're going to click on it. You get to see exactly what is being instructed here. And this is an interesting one because we're just having clot basically put on a mask and become a CMO. So sometimes it's just Claude acting as someone and not necessarily executing something. But as you can see a lot of information here, you can just copy this directly, maybe save it for later, add it to your own Github. So this is another example. If you go back to Google, there's a lot of other websites. A lot of reedit users in slash Clod AI do share their skills. Maybe you can put in your skills, get feedback on it, and there's a lot of communities out there that are helpful for this case. So my suggestion is GitHub, red it, for the most part, because it's people like you sharing the skills or sharing things that they found interesting. So my suggestions are Red it, GitHub, because those are people just like you who are trying to learn Cloud or are experts and want to share their expertise. Go ahead and use these platforms, refine your skills, always go back to them, try something new, and see how you can implement it in your workflow. In the next lesson, we're going to learn about connectors, which is another very important feature in Claude. Mm hm. 10. Introduction to Connectors: Connectors in clot can take the way you use clot to a whole different level because they allow you to automate your workflows and get things done within minutes. So you can think of workflows that you do every single day. Doesn't matter what you do as you work or if you're a student, anything from checking emails, checking slack or any other communication platform, checking your calendar, writing a to do list. These things take time, and you can kind of put that weight onto Cloud to basically act as your assistant. So in this lesson and the ones coming after, we're going to learn all about connectors, how to use them, what they're for. Some examples of workflows you can try next and how you get to apply it and combine it with the skills that we already learned about. So we have connectors and plug ins. These two are a little different from each other. They're not exactly the same idea, although most people overlook plug ins. I'll get into that in a second. But connectors are basically external applications that you get to connect to Cloud, give it permission and have Cloud do certain things with have some connected here. We have Gmail, Google Calendar, and Apithy and there are some suggestions up here for Slack Notion, Google Drive, and you can search for any other app that you use every day. The only requirement is that there needs to be a connection to clot. So for example, connecting it to Google Images. Your Google Photos isn't really going to help because you don't really do anything with those photos. Whereas in Google Drive, you have the option of sorting files, sharing it with everyone, adding other documents in the folder, and more actions for clot to take over. You can basically browse the connectors that are available in Cloud. So there's tons of them right here. You know, we have Air tables app here, even Spotify, there's so many different things that you could do. But if you're not seeing the app that you use every day, you have the option to add it by yourself, so as a custom connector. Now, this one is going to be a little different because you have to find a Cloud custom Connector page in that app that you're trying to add. So you put in the name, and you have to find the remote MCP server URL. So that's something you get from that application's page. So you would have to look into that. We also have these additional settings, but most likely if it's an app that everyone uses, it's probably available in the browse connectors. So let's take a look at what sort of connectors we have here and why we should even connect them. So if you go to filter B, you can see the different categories. If you want to do something with communication, clicking this will separate those for you. We have fireflies, even. We have agent mail, these things that you may recognize depending on your current but there are other things such as creative, for example, FIGMA. That's another tool everyone uses. Shutterstock is here, 11 labs. These are other AI tools, and basically connecting them allows the output of Cloud to be more tailored towards the FIGMA style files, 11 style files, and so on forth. We even have CFA available right here. So we have creative. If you do data and analytics, you can come over here, healthcare, education, even if you're a student, General productivity, if you just want to sort your task together, if you do nonprofit, legal, media, and a lot more. You can also sort these by what's popular right now, which is usually the best performing connectors. But we also have those that are trending. So these guys suddenly got a spike in use. So even Angel list is here for entrepreneurial work, and then the other stuff that you can sort via the alphabet. Now, if your app is not, like, popular, doesn't mean that it's not going to work. Means that there isn't much experience that Claude has with that app yet. So, like I said, there's a lot here. We have international apps and apps that are for one country. Like, there's a bunch of apps in Japan that necessarily people around the world may not know about. But if you're from a certain country and you're used to one app, try searching for it and see if it's here. You may be surprised. So those are the connectors, the external apps that you connect to Cloud, but we also have plug ins that are made by Anthropic so the company that made Cloud. And they're in different categories. We have design, marketing, and if you click on the Plus, they get added to your application list. But we also have some by partners. So this is a little bit different from connectors because Connectors is only the app sus Plugins, there's, like, some actions that Cloud is supposed to take when this is enabled. So let's take a look at an example. We're going to look at FIGMA so if you click on it, this is what Cloud can do. So you get to code smarter with Code Connect. You get to extract design context from the layer. So if you have a file in figma, Cloud can extract that. Cloud can look at your flows and turn it into diagrams, output it in FIC Jam for you. And, you know, these are the tools that is going to be applied here. And this is developed by FIGMa itself. This is another thing. Make sure you're looking at who made this connector. If it's a random person claiming that they made this FIGMA connector, I would double check that just for safety reasons because you do have to sign into your own personal account, and therefore, you need to take some measures for that. So make sure if you're trying to use the real FIGMa, it's developed by FIGMA. That's a connector. But if you go to plug Is, I just saw FIGMA somewhere. Go to Partners. There we go. This has more skills integrated. We learned about skills. So these are different instructions all put together in addition to the connector FIGMA. You can see connectors FIGMA then we have all this forward slash skills. So this is like a level higher than a regular connector. I'm just going to install this, for example, just so we can see what these skills are for and going to go manage and these are all the skills. So connector is only one, and then the skill let's say FIC Jim, we click on it. This is another skilled at MD file, and these are the unique instructions that Clot needs to take in regards to this specific connector. So adding images, what it should do. Only supportive way to add images. Do not use these scripts, avoid this variable bunch of rules that make sure that you get the best output possible that works with Figma. So you can combine connectors and skills to make plugins, takes it to a whole new level. And yeah, let's take a look at some other plug in. It go to Browse, we'll take a look at a few that are bienthropic. Let's try the productivity one. You can click on it to see what it entails. But basically, when you install this, you can ask for these suggestions, and it's going to tell you exactly, you know, based on what you have connected, you'll pull in information from maybe your slack, your emails to see what's on your plate today. If I click on this, it just puts the prompt in for me. And as you can see, it goes to cow. So if I have my connectors enabled here, it can pull information from these applications. Let's go back here and see what is in here exactly. So these are the skills. We can see task management, for example, file location. This is another skill, and we have a bunch of things that are being told to clot to do. So similar to skills, you can also make your own plugin, which is a very cool feature. If you go to add, you can get some from the marketplace. There's some Benthropic. There's some from a repository. So if you have a Github page that you really like, you can add that repository here and, you know, try those scripts. But if you click on that, we have some additional source this is really depending on what you do with Cloud. If you're in the legal department, you can add this sciences, financial services, and whatnot. But I'm going to keep the knowledge work because this applies to everyone, but feel free to add these if these are they can be helpful for your workflows. But other than that, we get to create plugins with Cloud or upload some we see online. The same places where I showed you in the previous lesson, you can download plug ins from there as well. You simply search Cloud Plugins, free or paid. If you can even put, keywords that help your search, download them, upload them here. Creating with Cloud we'll put in this prompt for you, and now you get to, similar to the skilled chat with Cloud, put in as many information as you want or as little. And make a plug in yourself. So that's going to be mostly in the cowork feature. And once again, to enable a connector or a plug in, you simply have to go over here, make sure it's turned on, and the same thing goes for the plugins. So if you don't see it here, you probably deleted it or it's just not running properly. So if this is the case, just turn it on. Sometimes there are updates as well. But they will all be listed right here. So that's basically what connectors are. I also showed you plug ins. They kind of go hand in hand, but we're going to move further to the next lesson and see what else we could do with connectors. First of all, how can you even connect things to Claude? Which one should you do? And what are some examples of common workflows that you can find helpful no matter what your job is? 11. Connecting External Apps: Mm hmm To connect external apps, there isn't much to do. The first thing you need is to make sure you have an account with that app. For example, if you want to connect Notion, just go on there, make a free account, and you're going to have to connect it to Cloud, which I'll show you right now. So for this lesson and the ones after it, I will be using the following external apps. Feel free to add more or disregard some of the apps that I'm going to mention. But I'm going to be working with Gmail, Notion, Giamma and Google Calendar. So four apps, but again, feel free to manipulate that to your needs. The first thing you need to do is go to customize right over here, go down to Connectors and click on AD. If you see it right here, then you can just click on it, but I'll do the long way to browse Connectors, and I'm going to add Notion. So I already added Gmail as you saw, but here's Notion. You click on the Plus. To read about it, again, make sure it's developed by that application, and then click on Connect. So when you click on that, it's going to launch a browser tab where you have to grant permission from that account. If you signed in with your Google, that's going to be your Google account, and you just have to make sure that everything goes smoothly on that side. So I'm going to do that with my account and then we'll be back so we can continue. Just connected notion. You can see we get this message, and now we get to click away. So now instead of connect, it says disconnect. And that's it. That's how you can connect applications. We can do another one together, Gamma. So if you're not familiar with Gamma, it creates presentations for you. If you do a lot of pitch decks, data analysis, you get to use this. It's really cool. Again, you click on Plus and then you're going to have to sign in here you're going to authenticate and then go back to Clock. There we go. Connected. Click away. And now we should have a few connected. So let's see. If I close this, there we go. We have Gamma, Gmail, Google Calendar, notion, and Apifi. But I'm going to leave this out and just make sure you have the apps that you're comfortable with Claude using and apps that you're going to use. Now if you are worried about permissions, Claude lets you adjust that to each of these applications. For example, Gamma right here, I just clicked on it, and I was brought to this page where I get to work with the tool permissions. You can decide what Cloud can and can't do. If you need everything to need approval, which I recommend, you can click on this kind. You can block certain things, allow certain things, and, you know, go about it as you see fit. And I highly recommend if these are the accounts that you use on a daily basis, meaning that you are using it for your work actively, that you go into each of these and make sure that you're aware of the permissions. So I'm going to, you know, make everything approval. Now, I do recommend that you leave everything for starters on need approval because sometimes Claude may execute certain tasks without you knowing. And if you do use that account on a daily basis, that may cause some confusion with your coworkers or just your overall image. So, you can also set it to manual here. When it's manual, Claude is going to ask you for permission about everything it wants to do with one of these apps, and that way you're just safer. So we have everything set up. The only downside for that is that if you have a scheduled task, which is going to be things Claude does at a certain time on a schedule, it may run into some issues because say it's like two in the morning, you're sleeping, Claude has been scheduled to do something because of time zone differences, you are going to run into some issues there. So just bear that in mind. That is fully dependent on what task you're trying to execute here as well. Alright, so let's go back to one of the chats. I'll just make a new one. We'll do cow. You can add a project here. So if you want to keep things organized, you get to do it here or just leave it out completely. Again, you can set this on manual and choose your model, cowork, we're ready to go. Now, with cow because there's multiple things you can have Clot do with one prompt. We get to combine Skills, connectors, plugins, together, even attached files to have Clot help us out with our daily work. So let's take a look at some examples. We're going to do a fake business together, and then we can see the full application of these connectors live. So let's make a new chat, and I'm actually going to do a project. So I'll just do a new project. Let's call this design trends research. So we're going to say a little bit about our goal, learn the trends on UIUX design and help with project management, create the project. And we're going to set a very basic instruction. So act as my market researcher. And manager provide me with accurate data points online and help me plan out my projects. So you can give it more context here. So let's say I'm a UIUX designer, and I work as a freelancer. I want to stay up to date on the latest design trends and be able to work on my web projects more efficiently. Set some rules that really helps. So rules only look at recent data, so maybe nothing before June 2026. We can say or actually, let's say, January. So for this year, include design trends relevant, let's say, to the fintech industry, provide detailed and actionable plans for me to then execute avoids online communities such as Redit or Facebook to do some rules here, give it some context, and just save that instruction. Now, we can have Claude do a little research. I'm just going to go over here and make sure the connectors I need is turned on. So we can do a Gamma one. I don't need Gmail here, don't need Google Calendar. Let's actually set up Google Calendar. So Gamma, Google Calendar, Notion. That's all we need. And I'm just going to, you know, ask it some questions. So first of all, I have a client that needs a website for their cryptocurrency business. Look into latest trends or maybe latest web design trends for the industry and provide me with the bullet points. One more thing I want to add here is the output format. Let's add a few more rules here, so keep summaries up to five sentences long because I don't want to spend that much time reading highlights the data points with bolded format, and that's it. So let's save that and provide me with a report. I'm just going to switch that out. So we're going to hit Sent. Right now, there's no connectors for research purposes, but if you did want to try that out, it's going to be something like perplexity, you know, Google Scholar, those sort of places. See if your favorite research tool is available and you can easily add it as a connector. So you can see on the right side, there's a bunch of processes it's doing, and you need to adhere to my rules, and you can see that it has its own connector right now, which is the web search. So if you're asking for something generic like research, Claude does have its own connectors that's not visible on the list where you download it or connect it. So either way, whether you have that connector or not, sometimes Claude does have its own local replacement. So this guy is running some commands. It may take a while, but I'm going to let this run and while that's happening, we can think about the output. So once I get this report, what is my usual workflow for, you know, my daily work? Say, I usually spend 5 hours researching trends, and later on, I had to make a presentation, maybe forward it to a few coworkers, made a to do list on my notion. You need to pinpoint these checkpoints in your workflow and then be able to iterate it to that is going to be our next step. Another suggestion that Claude is doing right now is giving us is making scheduled tasks based on this thing. So if I do research maybe every day or every two days, I could have a scheduled research session for Claude, where it does the research and just provides me the report ready to use. And that's a really cool thing that we get to do here actually. So we got the report right now. We're going to continue this in the next lesson because we need to see how we get to apply this report in relation to the connectors that we installed at the start of this lesson. We're also going to look at the schedule task and how we get to combine skills, connectors, and plugins to basically let Claude do the repetitive task on our behalf. So let's go ahead and see how we could do that. 12. Using Connected Data Effectively: So we have our report ready. I can see a little summary, but there is a document here that we get to take a closer look at. So we have some bolded information. The summary isn't that long, as you can see, and it's nicely separated into different sections, which is something really cool. I did not ask for that. This is probably Claude's own research skill and the format that it specifies. So I could, you know, use this now to do certain things. So we're going to use the connectors here. First thing I want to do is make it to do list, have Claude look at this report, and tell me exactly what I need to do for the next week to prepare my website for the client. And then we're going to have that checklist go into Notion, where I could sign in for my phone and keep track of all the stuff that I'm doing in a more effective way. So I'm saving myself the time of reading all of this, writing down the different tasks, and then putting them into Notion. So that's the first application. So I'm going to say here, based on the report, make a to do list. For me, make a to do list for the coming week, so I can have the website ready by, let's say, July 31, make these actionable items only. So we're going to get a very simple list, but we get to apply it into non using our connector. There we go. So today is the 23rd. Showing us exactly everything that needs to be done. Now, we can see that there's also some dates over here. So I'm thinking I'll have load estimate how long each of these are going to take. So modify the list with the estimated or maybe with an estimate of how long each task will take. And yeah, so maybe we could translate this estimated time into a Google calendar event. We got an hour, 30 minutes, 1.5 hour, and then so on forth. So now we have 32 to 34 hours across nine days. So let's see how we could apply this. So make a new notion page with the list and create these events on Google calendar. And I'm not going to say anything else. I just want to make sure that Google calendar is ready, Notion is ready, and then the Gammas, well, so let's actually say, turn the report into a nice presentation with Gamma. So three things it has to do. Let's see how it's going to work out. We have manual set up, and I'm doing this within the same project. So it's going to use the context as we go on. Let's hit Send. So even this guy turned into a little icon. This is just one application. Now, I did ask for a bunch of stuff, so it may take a while, but depending on how big your prompt and your ask is, it may take less time. Now, it's going to find the tools and create the task list. You can see that we have a few additional things here. We should see the connectors down here, currently, it only has web search. We should see the other connectors over here as it analyzes and goes forward. So Notion just popped up, create pages in Markdown. Got a little script, and this is where the manual part comes in. It's asking me for permission. And I'm just going to say yes, but this way, I know exactly what Claude is doing on my behalf on my notion account. We also have Gamma here. It's been called, and there is Google Calendar. So we're getting a bunch of other permissions. Let's I'm going to allow everything because the accounts I connected are not my main accounts. So I just had to reconnect my Google calendar. There we go. Now it's all running, and it's done. So two of three are done. One is blocked. Let's see what it's trying to do here. So, yeah, I didn't have one of the permissions enabled, but I'll just tell it that I just connected it. Here's our Gamma presentation. I made it pretty. I could edit this in Gamma but at least every information was brought over to my account. So let's say Google calendar is good to go so you could retry the command, and we're getting a bunch of stuff being executed. So once Google calendar is done, we get to, you know, go in, sign in to those web pages, see if it actually did create these events and, you know, make changes there if needed. So now we have the events there that Gamma deck is generating. So I think this is I'm not finished yet. Let's log into the web pages on our browser, Notion, Google calendar, and Gamma and see what popped up from Claude. So the first is Google Calendar. I just signed in, and I can see that it made those events. If I turn this two week, I could see the rest, as you can see, the times are set in properly. If I go into each one, there's, like, a little description for me. And this way, I could allocate my moving forward because of this project. So I set from today, which is the 23rd till July 31, and we can see that nothing else is coming afterwards. So that is for Google calendar. Let's go to Notion. I could see the new page right here. It was generated. So now I have this checklist. This was generated by Claude, and I could come in here, share it with someone, or, you know, change a couple of things. But this definitely save time for me. Lastly, is my Gamma page. Let's go to home. There we have it. So there is my web design trends. We have all the different summaries, got some nice visuals. And as you recall, in the previous lessons, we did try to do a presentation with Claude only, and it struggled with the visuals. So Gamma takes care of that, and you don't have to go in and look for images to complete your presentation. I can share this with someone, make some changes. And with one prompt, I was able to execute three things because of connectors. Going back to Claude, you can make some changes. So I could have it completely delete all the events in Google Calendar. Let's say my client canceled, and I don't want to have to do that myself one by one. So client canceled, remove all the Google calendar events. Again, one prompt. It has the permissions. It should be able to remove it. And when we go back and refresh on Google Calendar, it should not be there at all. So there's our eighth task. We're going to give permission here, and as you can see, it's deleting them one by one. Alright. Now, all 20 events were removed. Let's go ahead and see if that is really the case. And there we go. It just disappeared. So I don't have these events here anymore, and everything is, you know, sorted and is up to date with my client, you know, interactions. Now we know how the connectors work in real time. We tried three different applications. Feel free to try some more, and you should be able to see them here once you do connected and install. Next lesson, we're going to see how we could apply this to schedule task because that way Claude can do certain things, even if we're not actively behind the computer. So let's see how that works. 13. Best Practices for Connected Workflows: So in the previous lesson, we looked at three connectors, Google calendar, Gamma, and Notion. Now we're going to apply that to a task that is going to be scheduled for a certain time. So we're going to just go over to scheduled, go to new task, set up manually. Again, you can chat with Claude if you're not sure how to set it up. Let's try to do this together. So let's say design trend search, and report. This is something I want done every, you know, once in a week, and I don't want to have to come in here prompt clot every single time. So in the description, let's talk a little bit about what we want done. So research on UI UX trends and generate a report. So here you get to be more specific. You can have this B in your project. Frequency, you could do weekly, let's say, 9:00 A.M. Every Monday. So that's when I wake up. I have my stuff sorted. Perfect time to sit down here and have a report ready for you so you can go about your day. Choose your model here. I'm going to go with Sun at five, and we're going to give a little bit of description here using the connectors. So we're going to name the connectors, basically. So conduct research on latest design trends and form a presentation with Gamma. That clearly visualizes all the data points and allows me to understand the trends at a glance. We're going to set this to safe. I do want to try this, actually. So let's do I'm going to set it to today. It's Thursday. We'll go to set this to 12 53. It's 12 51 right now, so it should, you know, get activated. Then hit safe. So it should run in about 2 minutes and it's going to use the tokens, but that's okay. We get to basically get a brief every time we want to start our week. It's pretty cool. Another cool thing is that you can basically combine certain skills. Say I want to report via Gamma, I want briefing of all the messages in my slack. I want all my email sorted. These stuff you can easily communicate to Claude that you need. And we have one more minute here. I'm going to show you how this works in case you were curious. You get to add this project or drag the context from design trends. I'm going to add it here. And we're basically going to get interviewed here. So it's pretty similar to all the other skill generators. So say I want a morning brief weekly, and it's going to formulate something so what time let's say Friday 5:00 P.M. What should the brief include? Let's say, calendar? It's going to load the tools it needs. And if you don't have it connected, it's going to ask you to do that. So here's the brief, basically the description of what Claude needs to do, the name of the schedule task, and that's it. You just hit schedule. So I now have something done using the help of Claude, I didn't have to type anything. I just had to know what I want. So it's 12 53 right now. Let's see if we get the thing. So as you can see, it started, it's doing the little checklist here. We got the context, and it's going to basically repeat what we just did in our project. So every time every week, Monday, 9:00 A.M. It's going to do this exact thing, look into sources, generate a report, format that into a nice presentation with Gamma, and it's going to be ready for me to use. Now, if you want to test something before you actually wait until that day, there's also a run now button. It does it for you right then and there. You get to see if it works well. If not, you can, you know, edit it, make some adjustments. You can delete it and even deactivate it so it doesn't run on schedule anymore. And as you can see, it disappeared from this list. So this is going to be the same thing. I'm not going to wait for it to run, but as you can see, it's something that happened because we scheduled it at 12:53. So I'm just going to deny this and just stop the schedule, the task, basically, let's deactivated and go back to our business. Some applications that you can explore after this lesson is things regarding productivity that we already looked at. You know, you got emails, calendars, notion, checklist, that sort of stuff. You can also do certain things. Say I have a team of designers working under me. I could have Claude look at my brief and basically divide that task across four of my work. Then send emails to those people on my behalf. So that's another application. I could have Claude look for the latest trends in Machu. We did that a different time, and it could look for Amazon links where I could get the ingredients, send those links in the slack and look out for any responses that were made towards me. These are all applications. There's so much you could do here, but all you have to do is think about your workflow, what takes time during your day and see how Claude can step in and take over. The most important thing is being aware of the permissions and making sure that everything is set up right so that you don't create or waste your tokens on things that are not even relevant. Go ahead and try that. We did this together. You could try this for your own business or your own projects and see how it comes out. In the next lesson, we're going to combine everything we've learned so far and make a project from start to finish. So we're going to be building a fully automated workflow using skills, projects, connectors, and even plug ins. And that way you will have a full hands on experience out of all the lessons that we have had so far. Mm 14. Planning Your Workflow: For our final project, we're going to be building a workflow from scratch. And the first thing you need to do is plan it out and figure out what problem you're actually solving using Cloud. So what I'm going to go for is a newsletter automation that I could do with, you know, trending books out there, and I could have Cloud look into what books are at the top and exclude some things that I don't want to see and then put that into a newsletter where it's going to send it to different mailing lists on my Gmail. That way, we can combine a few things and also have Claude use Gmail, which is our connector for this project. We're also going to be making some skills as well as setting up our project where everything can be within one space, and Claude can draw more context as we go on. The first thing we need to do is try to figure out how we want to approach this. Do we need any plug ins? Do we need to make any skills? How do we want to set up the project and other things? I will be building an artifact during this process as well, just so I'm able to visualize the trending books in a way that I want to see it, and then it goes out to the connector, which is Gmail. So first, let's make a project, as always, going to make one right here. We'll call this book recommender. There we go. I'm not going to put in any context here, but let's give it instruction. So we're going to go for act as a book critic and compile a list of books that are very popular right now and then make that or turn that list a engaging newsletter for my mailing list. So context, we're going to provide some of that. I'm a let's say influencer that does a lot of book content, and I send weekly newsletters to my audience on Instagram and email, where I recommend new books for, let's say, my audience includes people aged between 16, maybe 18 until 45-years-old. They all speak English or only English, I guess. And want fun reads that are short, yet engaging. So then we're going to set a couple of rules here. Only look at the New York Times bestseller list. Avoid any other website for books include only English language books. Translations are okay. Avoid genres in fantasy, Sci Fi, and let's say, graphic novels. So we're just setting some parameters here and constraints. And then the output is going to be, let's say, one engaging Newsletter with a catchy subject line include only three in each newsletter. Provide a brief one to two sentence description about the book. Let's say we want to word this in a way where people want to click immediately. Say something that makes the reader want the book immediately. Provide the places that the book can be bought and their price. And let's say generate two Instagram post ideas using these findings. Generate good captions for the Instagram posts. This is how I'm going to set up my book recommender. We're going to save the instruction. You could put in more context here. If, for example, you have a certain way of writing your captions. This is a great place to upload it because Claude can kind of look at that and basically replicate the way you speak. And then one more thing in the rules, use warm and friendly tone of language. And we're gonna head safe. First thing we need is going to be a skill that can, you know, replicate my tone. Basically, I want Claude to understand the way I speak so that it could be consistent throughout the entire thing. So we're going to head over to customize skills, and we're going to create one with Claude. And then, I guess, I'll just send this. This came from Claude itself. So then I kept in this prompt, learn my tone of voice slash speech that you can replicate it for the newsletter or just for a newsletter and Instagram post. Ask me questions that would help you understand. My preferences provide different style for me to choose from. So very straightforward. This ended up creating a project, which is fine, but we're just going to see how we can build a unique skill for this workflow. So we have multiple things happening right now going to open these up so you can see what's going on. And now it's going to interview me so that we could do a better job with a newsletter. So should the skill save your voice profile to a file after the first interview? I think save it to a profile. When you say provide different styles to choose from, multiple drafts, same voice, just so I could see how it's implementing this. What should the skill actually produce when you ask it to write something voice guide only? That's a skill that I'm going to put in the project. So here I contradicted myself a little bit, so I ended up saying voice guy plus draft posts. This is something I can also put in the file if needed. So it's been a few seconds, and you can see that it's writing the file right now. And because I've been speaking to Claude this whole time, it kind of has an idea of my way of talking. So if it didn't ask you a lot of sample writing, don't be surprised because you've already talked a lot with Claude and it knows exactly how you form sentences, what sort of words you use and other stuff. We have our skill dot d file. It's going to now draft files, and at some point, it's going to ask us for feedback, so it's going to be a fully tailored file that we can look at and then download to bring into our project. We even have a template right here, which is going to go within the skill. All right, so it drafted the skill, and it has three test cases, and it just wants to check in with me and see if it's going the right path. And this is going to now run a few commands. Now, while this is happening, you can think about the tools that you want for flow. As I said earlier, mine is going to include Gmail, but I also have Instagram post enabled in my project. So that's going to call for Canva, something that I could, visually design with. And then, I just downloaded from Canva posted onto my Instagram. So we're going to set that up right now while this is happening. So let's go ahead and move on to customize the move down to Connectors. And we're just going to add Canva. Should be right up here. So there we go. I just connected my Canva, and we're going to jump back into Clo now we have camera ready, all set. So now we have six agents running for this skill, and we can see the files being put out here. This is just Claude process. It's completely normal. But going to the project. So if you're not seeing it here, just go to Project, and it's going to be right here. We want to make sure that we have for this project, we need Canva and Gmail, so I don't need these guys anymore. There we go. And I could have that modified into my project right now, actually. So in the instruction, output, generate two Instagram posts with Canva. I need to name it here. So, let's say, we'll put the newsletter in Gmail. Draft newsletter in Gmail, with the subject line, but this after this line. So it's newsletter than Instagram post. Okay, so that's running. Let's check on the skill. It's at step five, almost there, getting a bunch of output files right now. So now it basically made an artifact for us. It's in HTML, and we get to provide this feedback into clot. So these are three different cases. So let's read this together. W the front on the browser to review three test cases side by side with the skill and without for each one, leave feedback in a text box, then hit submit all reviews. So I'm supposed to read these. So the prompt with the skill provided, and then see if I like the way clot takes that text and applies that scale to. The user is talking like this, and then Claude is going to extract a few things. So it's like a simulated conversation between Claude and the user. Let's say I agree with the way they're speaking, I'm just going to go with sounds great. Keep it warm. Next, we have one without the skill applied. So you can read all this and then see if there's any errors you have. Let's say this is less warm than the previous one. Next, this is, again, with the skill. Looks great. So I'm not going to read the whole thing, but the way you spoke in those newsletters were very important. This is a way to tailor the voice with Claude. So I just downloaded the JSON file. I'm going to upload it again here so Claude can work and continue with its different steps. So the same file that I just downloaded, upload it right here, so Claude can continue. Also put in and provide Claude with a sample text yourself so that it's not, generating something on its own. But if you don't have anything like that, as you saw, we were able to do the same thing. Okay, a lot of waiting. We're at the last step right now. There we go. So there is my skill. Have something called voice profile, and I get to save this skill into my list. If you want to make changes, you do it now, and then you update the skill. Now, to implement this in the project, we're just going to go back here, book recommender instruction, and then we're going to say use at voice profile or just tone save Instruction. And we get to call it here as well with skill. So let me see where it is. There we go. Now it's actually built in for this project. So, we were able to set up a workflow. Let's see how we can connect the tools and run a few tests together with our book recommender. Mm 15. Connecting Your Tools: Alright, so far we have Canva, we have Gmail Connected. Let's take a look at some plug ins and see what would work for this scenario. So we're going to head over to customize, then go down to Plugins. We made a skill, we made some connections. For Plugins, I'm looking for something in the writing realm, maybe research, something. I think marketing works well here because email marketing is relevant. Have let's see what else do we have? Let me go to Partners and see what we have brand voice. That's something we get to implement here. So for Zoom. And I think that's it. Let me see if there's one for Canva. Okay, there we go. I'm just going to get one for Canva. Gmail probably doesn't need one, yeah. Alright, so we have our plug ins, connectors, and we have one scale. Let's go back to our project. Book recommender, and we're just going to enable a few things. So we have came just to make sure we're on the same page. Canva, brand voice, marketing, and the rest are not really needed here. But connectors, make sure you have Gmail and Canva. So let's say generate newsletter and Instagram post for, let's say, June 2026 to the last month. We're going to go in cowork and start the task. Let's see how Claude approaches this and what sort of questions it's going to ask us and, you know, see what we can do here. Now, there's a tip down here, which is really interesting and we could try it out. That's going to be for like a dashboard for us, which I think would really help in this case. So let's send this as a cowork so multiple tasks being done at the same time. You get to do that right here. This is within the same project. So it's going to use the same context. So we can see that it looked at New York Times Best Sellers for 2026. You can see it's asking for some things. Right here, what does it need? So our dashboard dot HDML is being built right now, this is going to be a live thing, and I get to add this to a project, actually. Let's put it in book recommender. I'm just going to go back and forth and see how each one is doing. We did ask for a large a lot of work for cloud to do, but this is just one application of a good workflow. Now it's asking me some questions. I did say that it needs to use the voice profile skill, but these are some things I didn't really specify in the skill, so we're just going to answer this real quick. So newsletter could be straight to the point, and then Instagram can be trendy. Number two, no emojis. Number three, for sign off, we can do sign off with Have a great day. Number four, we never wanted to sound. I guess we could say too, yeah, too salesy for sure. Any newsletter or IG accounts we want to replicate here when I say no. So that was a good interview. In the personal dashboard, we're getting a sample here of, like, the different events. We're getting different emails here and upcoming events. If I was a professional book recommender, this would really help. It's live. And if you press on this, it becomes its own webpage. The Gmail and the upcoming events is disconnected. It's okay. But you can kind of play around with the second panel track, but I'm going to just pause this right now because our focus is on the actual newsletter. Alright, so this is our draft. Claude is building it right now. Draft newsletter, that's the step that we're on. And you can see that it ended up using the voice profile skill that we just made, and it's using the connector Gmail so far. So let's allow this task. So I'm going to let this run. Again, it may take longer, but we're going to come back later on with the way it looks because it needs to actually visualize the things on Canva. So I'm just going to let this do its thing, and then we'll be back in the next lesson to see what Claude did, how it kind of orchestrated this whole workflow, and what else we could do to make this better. 16. Final Review & Next Steps: So this ran for a bit. We now have the Canva design as well as the newsletter draft. So after my answers, we have a few designs for, like, the general newsletter picture. And then we have more designs regarding the book itself. So this is a book that it picked out from the New York Times best seller. So the Klamty Club. It's right over here. It has some pictures for us, some visuals, and all I have to do is, you know, download the one I L. Now, it also gave me a bunch of sources for pricing. So we have Barnes and Noble and Horizon Books. We also have the Instagram captions that I could, you know, look at, and it even linked the design for me. So the spotlight for the book, maybe the Book of the Month, this is the design. So this is the design for the Book of the Month. It was linked in the document. So, you know, I just have it downloaded here. I want to make some changes, you know, Canva has a lot of options there. And the same thing goes for the other post, which is three books to read this month, like a roundup post. Going to open this up, and this is our other post. Looks pretty good. Then we have our newsletter. So let's actually open our Gmail drafts and see what it wrote. I did not have it send it to anyone just because we want to make sure, but you can also attach an Excel sheet with your mailing list to make this a more automated process. Here is the draft, as you can see, we have a nice subject line. We have the three books, where we can get it and the pricing. And then it closed off with what I mentioned. Have a great day. And this is going to be repeated every time you ask for a new round of newsletter. So maybe you could try this next month and the month after. And as we know, you can also do a scheduled task, such as generating the book reports every month so that you don't have to go in here and prompting it yourself. You looked at that, but again, it's in scheduled. You can make a new task with Claude or yourself, call it Best Book Research or something, have that run on its own while you have this project to do more of an in depth application. So going back here, we so far had a bunch of steps that were done. Have voice profile, June 2026 captions, all within this project with these connectors. And we also have the skill that we made at the start. So now I could restart this and have it do this process for maybe next month. I'm not sure how often New York Times bestseller gets updated. So we could do something else now, now compile a list. An output for the best Romance books. So, again, it needs to avoid sci fi. It needs to avoid fantasy and only English books. So it's just double checking a few things. I said the full newsletter and Instagram post, and then New York Times list only. Feel free to make changes, but we already have the setup here, so it should not take as long as the first time that it did this. Also got a bunch of stuff being looked at over here. It's looking at certain sites. So you can look at certain sites, if you want. But I only said New York Times best seller. Fetching URL and all of that. I'm just going to collapse this guy. So we have this going for the schedule task, you can incorporate the project just so that it could drag in the context from there. So you just attach it like that, and it's going to use that information. So let's go over here. It's asking for permission. Yes, it's doing the Canvas style right now, designing them since Canva has AI itself, it's using that capability to make the visuals for you. So that's a great benefit of making your connectors before you start the project because Clot could just work with it independently once you give it permission to make you some nice visuals. So you can see it's pretty similar to the previous Canva designs, and that's how it keeps its consistency. But you guys can upload some stuff in the context box right over here, maybe your own personal branding design. Even if you have a brand guide from a graphic designer, you can upload that PDF here. So here's my other set puts, we have two posts for romance books, and then we have the newsletter drafted in Gmail. So these are saved in Canva, and then we have the captions right over here. Again, it's telling me where to buy these books, and this will continue to execute as long as you have the instruction perfectly laid out. That is the end of our workflow. We were able to build something together from scratch, pinpointing a repetitive task that we had to do. In my case, it was writing newsletters and doing research. In your case, it could be something completely different. From now on, the more you use this, you get to make some changes and improve the instruction file and even make some additional skills just to make this experience a lot more efficient for yourself. So we already saw how to make skills, feel free to make more, add more connectors if needed. And just like this system that we designed, you can apply it to any other field that you're in. So just experiment with it and see what you can build from now on. 17. Class Project: Build a Business Claude Workflow: Now it's time to combine everything that we have learned into building a reusable clod business workflow. For your class project, you're going to select a reoccurring task that happens within your work, your studies, or your everyday life. Once you have that selected, you're going to start defining your workflow. So that means giving clod the information it needs, the steps needed to complete that task, and what sort of constraints you need to give once you have that, you're going to create a project dedicated for this task. You're going to begin by writing Claude and instruction that it could use to better understand the context, the steps you need to take and if there's any reference materials it needs to use. Then you're going to create at least one reusable skill that you can use throughout this process. Once again, you're going to give it a clear instruction and have something built that you could apply to various inputs. Also use a cloud connector if your plan allows it. Connect an external source that is going to aid your workflow, and it's something that you also use every day. Just make sure that the connected data has a clear purpose so nothing gets lost throughout this project. Finally, you're going to run this project from start to finish, give it an input, have it do the steps that you provided, and watch the output. Is the output useful? Can it be better? What else can you do to take it a level your project submission, you're going to be combining the following things. These could be in screenshots in a diagram that you can draw or a written overview. The first thing is the problem that you solved. The second is how your project is organized. The third one is which skill you created that was reusable. The fourth one is any connectors that you used, and lastly, what changed after testing. If you did not change anything within your project after testing it, you can also write that down. Once your project is complete, upload it to the class project and write a little bit about your experience, what you found difficult, or any other information that would help your fellow classmates understand your process. Good luck with the project, and I look forward to seeing what you've made. 18. Congratulations! What’s Next?: Congratulations on completing the clod AI workflows master class. Throughout this course, you went beyond the basics of clot and learned how to build reusable systems using all of its unique features. Remember that strong AI Skills comes with practice and with time. It's never perfect on the first attempt. So try building more as you go on, try different things within the project that we did and see what else you could discover using this very amazing tool. You haven't already, be sure to upload your submission to the Project Gallery. That way you and the other learners can understand how you approached the project, and what was it that you tried to solve using Claude? If you found the course to be helpful, please leave us a review as it really helps us better our services and see what else you would like to see in order to improve your learning journey. Thank you for joining me and good luck with your Claude journey.