AI for MARKETING AND BUSINESS: ChatGPT Masterclass: Learn How To Use ChatGPT 5 Effectively

1. Introduction: Guys, and congratulations on taking this class, ChatGPT master class, how to use ChatGPT effectively. If you've ever opened ChatGPT, type something in and thought, this is good, but I'm sure I could be using it more effectively, you're in exactly the right place because ChatGPT isn't just an AI chatbot. It's a full productivity system. If you know how to use it properly. In this course, we're going to take you step by step through how to use the newest current version of ChatGPT effectively, and how to apply it confidently to real work, real projects, and real decisions. Let's start with a real life scenario. Imagine you're running the day to day operations of a business or in any busy role where you're constantly switching between tasks. You're researching topics, writing emails and documents, analyzing information, responding to messages, organizing ideas, creating content, chatting accuracy, and trying to keep everything consistent. You can use ChatGPT for all of most people don't because they treat it like Google or a basic chat box instead of the powerful workspace it is. In this master class, you'll learn exactly how to use ChatGPT the way it was designed to be used. You'll learn how to create consistent repeatable text using the right structure and constraints, generate high quality images, and refine them for real world use, turn rough notes, uploaded files, and ideas into polished professional documents. Reduce repetition, hallucinations and setup time by using ChatGPT memory strategically, perform data analysis to spot trends, patterns, and insights. Duck fast reliable topic research using built in web search and deep research, organize ongoing work using projects instead of scattered chats and combine all that knowledge to create a custom GPT that will act as your agent and assistant to handle your queries for you 24 hours a day, seven days a week. With all of this, you'll know how to turn hattPT into a reliable assistant for writing, research analysis, planning and decision making. Just as importantly, you'll learn when to use each feature and when not to. This course is about mastering ChatGPT to save you time, be more productive, and make you money way faster than your peers. You'll see live practical demos using the latest ChatGPT model at this time, so you always know where to click, what to use, and how to apply it immediately in your own work. The end of this course, you won't just be using ChatGPT. You'll understand how to think with it, structure work inside it, and use it super effectively like a boss. So take your time, follow along with the demos, and experiment as you go. And by the end of this course, you'll be using hatTUPT faster, smarter and more effectively than anyone else. So let's get started. 2. The Birth of AI and ChatGPT: So to understand why we even need prompt engineering, we need to understand where it came from and how it relates to AI, the birth of chat GPT, and LLMs. From the simple rule based programs of the 1980s to today's smart creative chatbots, one thing has stayed the same. Our goal to make computers understand and respond similar to humans. The story of chat, GPT, and large language models, LLMs is really the story of how that dream became real. How we went from basic machines to powerful tools that can think and write using everyday language. Although it feels new, artificial intelligence, AI has actually been around for decades. To understand where ChatGPT came from, let's take a quick look at how AI evolved over time. It all began in the 1950s when computer scientist Alan Turing asked a famous question. Can machines think? That one question started the entire field of AI. In the 1960s, a simple program called Eliza was created at MIT. Would hold short conversations by matching patterns in text. It wasn't truly intelligent, but it was the first step toward computers that could use language to communicate. In the 1980s, AI was used mostly for what were called expert systems, programs that followed rules written by humans. These systems could give medical advice, approved loans, or help design products. They were useful, but they had one big problem. Couldn't learn or adapt. If something changed, you had to rewrite the rules yourself. The 1990s and 2000 brought a big shift, the rise of machine learning. Instead of being told exactly what to do, computers started to learn from examples and data. This approach quietly powered many products we all use today. Spam filters learned which emails were junk. Google search got smarter at finding the right pages. Amazon began recommending products you might like while Google Maps learned to predict the fastest routes based on live traffic. Came alexa and other voice assistants, which could recognize speech and answer questions out loud, something that felt almost magical at the time. Computers were now learning from experience, not just following rules, but even then AI couldn't really create things. It could predict and categorize, but not write, imagine, or explain. That changed in 2017 when researchers at Google developed a new system called the transformer. It helped computers understand how words relate to each other in a sentence, not just one word at a time, but in full context. This was a huge breakthrough and laid the foundation for the next big step in AI, large language models. A large language model, LLM is an AI trained to understand and generate human language. It learns from massive amounts of text, books, articles, and online content by spotting patterns in how words and ideas connect. After the transformer architecture was introduced in 2017, models could finally understand context and meaning. This led to powerful systems like GPT, capable of writing, summarizing, reasoning and chatting in natural human like ways. That next step came from Open AI, the company behind ChatGPT. They built on Google's work and created something called the GPT series, Generative pre trained transformers. The first version GPT one showed that a computer could learn to write readable text by studying huge amounts of online data. Then came GPT two in 2019, which could write essays, stories, and even news articles that sounded human. A few years later, GPT three made an even bigger leap. With 175 billion parameters or neurons, it could write, translate, answer questions, and even code. There was just one thing missing, natural conversation. GPT three could give answers, but it couldn't chat as fluently. So Open AI improved it using feedback from real people, teaching it how to respond more naturally and politely. The result was chat GPT, a version that could hold a conversation. Remember what you said and reply in a way that felt personal. That was when AI truly became something everyone could use, and this is where prompting came in. The word prompt originally came from early computers. It was the line on the screen where you typed a command. The computer was waiting for your input. Over time, the meaning changed. Now, a prompt means the message or question. You give an AI a system like ChatGPT to tell it what you want. At first, prompts were very simple. Things like write an email, summarize this text or explain this to me like I'm five, but people soon notice something interesting. The way you wrote your prompt completely changed the answer. A detailed prompt gave a detailed result, a clear question, got a clearer answer. The better your prompt, the better the AI's performance. This turned the act of prompting into both an art and a science. Today we call this art and science prompt engineer. 3. Why you need Prompt Engineering: Okay, guys, after learning how chat GPT, and large language bbles came to life, the next step is understanding how to talk to them, and that's where prompt engineering comes in. If LLMs are the engines of AI, prompts are the steering wheel. They allow you to control the direction, quality, and creativity of what AI produces. At its core, prompt engineering means writing clear, structured instructions that guide an AI model like hat GPT, Claude, Gemini, or others. To produce useful, high quality results. A prompt could be as simple as write an email or as detailed as act as a marketing expert and design a campaign for my new product. The better you describe what you want, the better the AI's output will be. But how did this idea begin? Prompt engineering emerged as people started using AI tools like hat GPT, Mid journey, and Dali more creatively. Users quickly discovered that two people could ask the same question and get completely different answers. And the difference came down to how they phrase the prompt. This led researchers, creators, and educators to study the patterns behind effective prompting. Early thought leaders like Ethan Malik, Andre Carpathi and Seth Dobrin along with OpenAI's research teams began to share techniques that worked consistently. These evolved into the core prompt patterns used today, frameworks that help users think and write more strategically. Among them are the instruction pattern, giving clear direct demands, chain of thought pattern, guiding the model to reason step by step, persona pattern, assigning the AIA specific role or perspective, template pattern, creating reusable prompt structures and the iterative refinement pattern collaboratively improving the output. These patterns became the backbone of prompt engineering, a way to get predictable, powerful results from any AI system. Prompt engineering has since become the bridge between human creativity and machine intelligence. It's powerful because it gives anyone not just programmers, the ability to direct AI systems to perform complex or creative tasks. In short, it turns you into a kind of AI conductor, guiding the output you want, and it's not just for chat GPT. The same skill applies to a wide range of AI tools. Used across different industries, Canva, create marketing designs and social posts from simple text prompt, photoshop firefly, generate or edit high quality images using natural language, Notion, AI, draft reports, summarize notes, and automate documentation. Runway ML, turn written scene descriptions into professional videos. Microsoft 365 copilot, write emails, analyze spreadsheets, and prepare presentations from prompts. Prompting is now a universal AI skill. Whether you're a marketer, teacher, designer, developer, entrepreneur or project manager, learning to prompt effectively will help you work faster, automate tasks, and unlock new opportunities. It's also a practical way to earn more. At work, prompt engineering can save hours, automating emails, reports, and customer interactions. Entrepreneurs use it to scale content, analyze data, and generate business ideas without hiring large teens. Freelancers now sell AI based services from writing and design to strategy and automation. The people who know how to communicate with AI are already leading the way in productivity and creativity. Here's what I want you to do first. Imagine you've just been asked to use AI to make your current role faster, smarter, or more effective, or maybe even to help you land a new role entirely. Think about your day to day work. Are you managing projects, designing presentations, writing reports, planning lessons, selling products, or building marketing campaigns? We'll be using Chat EPT throughout this training to show you exactly how it's done. But remember, what you'll learn applies to all large language models and many of the other AI tools we mentioned earlier. Like Canva, Notion and Microsoft Cope either. The goal is for you to think in terms of how AI can assist you no matter your role or industry. If you're a project manager, imagine AI helping you summarize project updates, create meeting agendas, and identify risks in seconds. If you're a marketer, think about how it could draft, ad copy, analyze competitors, and plan social media posts, a teacher could use it to generate quizzes and lesson plans. A business analyst could interpret data and find patterns. A freelancer could create proposals, automate admin work and deliver projects faster to make it practical, is an example of someone running the day to day for an e commerce business because it applies to so many roles. So take a moment to think about how an AI, I can help you in your role because by the end of this course, you'll know exactly how to make that happen. So let's get started. 4. Getting Started with ChatGPT - Generating Emails: Hey, guys. Welcome to this section on how to generate emails with the instruction pattern. So what we're going to do today is we're going to get practical hands on experience with one of the simplest and most powerful prompt types in all of AI, and that's the instruction pattern. By the end, you'll know how to use this pattern to create high quality professional emails that save you hours of work that you can use, whether you're in marketing, sales, project management, or running your own business. Before we dive in, here's your quick guide. So first of all, the pattern the patterns called the instruction pattern. Why would you use the instruction pattern? We use it to tell AI exactly what we want to do clearly and directly. Its common uses are for things like emails, blog posts, reports, ad copy, and meeting notes. So the pattern is the foundation of Prompt Engineering. Silly allows you to give a clear instruction, and then AI follows it simple as that. No code, no setup, just plain language that gets results. So again, start a good chat GPT. The official way to do it is to go to chat dopopenai.com. So I'm going to do that. Official way. But as you can see, it redirects you to chatchpt.com, anyway. So I will just type in hattpt.com because it works. So now let's get started. And by the way, that works on the mobile as well. Now, why don't we get straight into the instruction pattern. So throughout the course, we'll be using one main real life scenario, and that's that you manage an Ecommerce T shirt brand, and you want to grow your business using AI. The same skills apply to nearly every job. Whether you're a marketer, project manager, business owner, or creative professional, you will be able to take these same ideas and apply them to your own role. So let's start with the exact first prompt using that we're going to use in our example. So you can type this in, but I'm going to be just grabbing them and putting them in. So the first one is write a sales email to sell T shirts to customers of our T shirt brand. So this is something a task that most people are going to have to do. So let's do that. So if we put that into ChatGPT let's see what happens. So that's a short, clear instruction, and it instantly works. And here you can see what Chat GPT has generated. It's allowed us to pick one of many subject lines, even including a Moji. Your new favorite T drop, fresh styles, perfect fit. Meet your next, go to shirt. Limited drop our best teas back in stop. So it seems pretty current. It said the word Ts in there, not T shirt. And the email body is as follows. Hey, first name, ready to upgrade your T game. Our latest collection from brand name is here, made with premium cotton bowl designs and a fit that keeps up with your every move. Whether you're turning on the weekend or out to make things happen, these Ts are built for comfort and style. Here's what's new. Super soft breathable fabric, fresh color ways and designs, ethically made, built to last, and for a limited time, enjoy X percent of free shipling all that good stuff when you order today. It's quite a lengthy email as well. Drop the collection now. Again, it's got a Mojis. Don't wait. Your size won't stick around for long, so it's giving you some urgency there about getting started. Stay called the brand name, the team website link, Instagram handle. As you can see, it's also asking us if we wanted to do anything more. Mike tailor the email for a specific audience segment, returning customers, first time buyers, or subscribers who haven't purchased yet. And it says, that would allow Chuck GPT to make the tone and offer more effective. So the first thing I want to draw your attention to is the fact that it's given you so much, and that's just from a simple instruction. So it clearly works. It's clearly current and uses current language emojis and all that good stuff. So that's a full marketing email ready to go, and that was in under 10 seconds. The next example write a friendly promotional email to customers introducing our new eco friendly T shirt collection made from recycled cotton. Keep it short and persuasive. Let's run. So, same thing, got a number of subject lines, big news, brand new eco collection is here. So you can see, this one was actually shorter and sweet. It's made that decision for us. So notice how we've added more context. The tone is friendly, it's product eco friendly collection, and the style is short and persuasive. Just notice those things about what it's done with this patent. So next, let's have a look at another So let's ask it to write a polite email apologizing to a customer for a delayed t shirt order and offer them a 10% discount on the next purchase. Let's see what it does for that. So here go. You've got a polished and empathetic apology email, so it's aware of emotion that you can send to your customer. Got a couple of subject lines. We're sorry your order is delayed, and our apologies for the delay here something to make it right. And then the email buddy, Hi first Nam, you want to sincerely apologize for the delay in delivering your recent t shirt order. We know how frustrating it can be to wait longer than expected, and we truly appreciate your patience. Our team has been working hard to get your order to you as quickly as possible and it should be on its way soon. As a thank you for your understanding, please enjoy 10% off your next purchase. We the card thank you ten at checkout. We value you as part of the brand name community and are committed to making experience with us a great one. Thank you again for your patience and support. Warm regards the brand name, the team, website link. Contact info. It's asking if we'd like to make it more casual or more formal depending on the brand's tart. So again, if I got that email, I'd be pretty convinced it was from a real person. It writes it just like a real person. I would It's been sincere and apologizing on our behalf. So this is definitely good to go something that we could trust to send to the public. And so in summary, here's what I want it to get across to you and here's what you've learned today. The instruction platin is one of the most direct and useful prompt types. You simply tell AI exactly what you want to do, and it follows your lead. You can use it to create sales emails, customer updates, reminders, and even internal messages, and other things that we talked about at the beginning. It says hours of work trying to work out what it is that you want to send in an email. It boosts consistency and lets you communicate clearly with customers and teams. So the structure to remember is the verb, the thing that you want it to do, the task, and the context, and any optional details outside of that. So as you move forward, start thinking about your own job or your own business. What messages or updates or follow ups do you send often? Try rewroting those tasks in the instruction pattern, and you'll be amazed at how quickly AI handles them for you, saves you time. So now it's your tone. Go ahead, and I'll see you in the next lesson. 5. How To Get Clear Answers Using Prompt Structure: Okay, everyone. In this lesson, we're going to learn the simple but powerful skill that's really going to make chat GPT instantly more useful at work, and that's prompt structure. The reason we're teaching that is because most people type a quick question or a quick instruction to chat GPT and hope that they're going to get the response that they want. However, that's why most people get vague answers and they get different formats, different responses that don't usually match what they actually need. So prompt structure fixes that issue. Instead of simply asking a question, you're going to learn how to brief Chat GPT properly, so it gives you a clear, usable answer on your first try. And that's how to be most effective with hatGPT. And we're going to use our real day to day scenario of running the daily operations for an ecommerce company, and we're launching a new track suit line. So, so here we're just getting started launching our new tracksuit. We're completely clueless about how we're going to do that. We're a new company. So we've come in, and we've said to ChatHPT, I want to launch a new track suit. What should I do? Just an open question. Let's see what ChaiBT tells us. So here we see that ChachiBTs come back with a lot of stuff. It says launching a track suit is a mix of product brand and execution. Here's a clear step by step roadmap you can actually follow from idea to first sales. Now, that's good. However, what you'll notice is, we've got a lot of stuff back here. We've got eight different steps here from defining the track suit, designing the samples, finding the right manufacturer. Branding it, pricing and margins, building hype before selling, launching, collecting feedback, and then it gives us some optional things that some optional things that it can do to help us even more. Now, while that's great, it's not very focused because in a company, sometimes we know exactly what we want to do, or sometimes these steps might change based on the goal, based on what we want to do in the end. So now let's add a little bit of structure to our question and see what happens there. So if we zoom in here and have a look, we've given it a goal to create a clear product goal for a new premium Traxit launch. The role, you are a product lead. So we're telling ChachPT it's the product lead so it knows what mindset to get into to do this, the context. The brand is a premium, modern athletic brand. We're targeting customers 25 to 40-years-old, and then for the style, it's going to be more of a performance based style. The key features are slim breathable, minimal branding, and the channel where we're going to be promoting this is on the online store and social ads. So we've been quite minimal in how we've worded this, but it's enough for Chat CHiPT to get the direction and the context within which we're marketing and promoting this product and the constraints. So we want the feedback practical and specific, no fluff and no longer explanations. The output format should be a product goal in one sentence and a success metrics using three bullets only and then some key decisions needed next, which should be five bullets only. So we've been really specific about what it should give back. And the reason if we're running a business, we know exactly what feedback we want. We may not need all of the steps that Chat GPT gave us before, but this will give us enough just to get the starting point for our product and we can go from there. So if we run that, let's see the difference between the output we got, which was an eight step plan and what we get back now. So here we can see we've got much more concise output. It's given us exactly what we've asked for. So a product and the product goal is to launch a premium slim fit, breathable track suit for style driven, performance focused adults age 25 to 40, achieving strong direct to consumer validation through online sales and paid social. So it's a lot more targeted via this goal, and it's something that we could explain to anyone in our business, the success metrics, and then the key decisions that we need next. So here we've got a product we know what success looks like, and we know what key decisions we need to make next. Now, if had we wanted a step by step plan, this would have been great what Chat GPT gave us. You can see already simply by being specific about what we wanted, we got a much more specific answer, and that allows us by being specific about what we wanted to get the starting point we want and avoid wasting time. The key thing to remember here is that the format you give it, the goal, the role, the context, any constraints, and the specific output format can completely change what you get back and be a much better use of your time when using Chat GPT, allow you to use it much more effectively. But in sum here's what we've covered. So we know that vague prompts produce vague answers. Prompt structure turns Chat chiefeet into a reliable work assistant, and that's what we've just showed. The core structure is the role, the goal, the contexts, any constraints and the output format, and we used it to generate a clear product goal for our track suit launch. So now it's your turn. So what I want you to do is I want you to pick something you need to do this week in your projects, in your business, in your career, your company, or in your personal life and write a structured prompt using the template. Tell Chat GPT you are a role, the goal, the exact outcome you want, the context, and only talk about what matters in that situation. So you can talk about the situation that you're in. You can talk about the environment that you're going to be in and anything that matters specifically to achieve this goal, the constraints, so the length of what you want back, the tone, and the rules around it, and the output format. Do you want numbered sections? Do you want bullets or a table? And examples of things you could do you could create a product goal. You could write a product description. You could draft a launch plan for something you're doing in different steps, or you could summarize some research, some competitive research into decisions. The key thing is once you get this right, Chat GPT stops being just a chat bot that gives you random information, and it becomes a tool that you could work through. So that's the end of the lesson. I hope that was helpful, and I'll see you in the next one. 6. How To Write In Your Brand Voice Using Custom Instructions: In this lesson, we're going to show you how to use Chat ChiPT to sound like your voice, say things in your voice and to sound like your brand by default. So most people prompt every time with things like write this in a professional tone or make this sound more confident or use our brand voice. And that works, but it's not the most efficient way to do it. What we're going to use today is something called custom instructions that allow you to set persistent writing rules so that automatically uses your tone, matches your brand personality, and formats responses in the way that you like. So the feature we're using is called custom instructions. While you use it to set a default style and behavior. What it sells, solls inconsistent writing, repeated prompting and off brand outputs and things that don't sound like your voice. So in order to find custom instructions, you'll look down here to your profile name. Click on that and we want to go to personalization. And then within personalization, we'll go down to here custom Instructions. So in this custom instructions area, this is where we can essentially give it additional behavior at home preferences. So as we're running an ecommerce brand, and this is for a track suit, we want everything to be based around that brand, and we want the brand to be really clean, premium, professional, confident, and we want everything that it returns to be based around that brand. So let's type in some instructions that will help it to do exactly that for us. The custom instructions that I've given it are we in a confident, modern and premium brand voice. The tone should be clear, professional, approachable, slightly aspirational. The writing style rules, short to medium length sentences, no hype or exaggeration, no emojis, avoid buzzwords and cliches, prefer clarity over cleverness, the formatting preferences, use headings, use bullet points for features or benefits, avoid excessive markdown and assume the content is for a modern apparel brand selling premium tracksuits. So this really shapes the way it's going to respond to us going forward. First of all, we're going to save, and it says, custom instructions are dated here. So we know that that's worked. Now if we click away from that and go back to our chat. Now we're just going to put in a simple instruction to write a product description for a black slim track suit and see how we're not saying anything more than that. It's just a simple one liner and it's now going to use the custom instructions to do this. So there you have it. Based on everything we've told you, it's it's related to a black, slim fit performance track suit designed for movement, refined for everyday wear. This slim fit track suit delivers a clean silhouette with the performance features you expect from modern athletic wear crafted from a bruvable lightweight fabric with built in stretch. It moves naturally through training, travel, and daily routines. The fit is streamlined without restrictive, offering structure where it matters and comfort where you need it. Minimal branding keeps the look sharp and versatile. Every detail is intentional from the tailored cut to the smooth finish. Key features, slim athletic and it goes through some of the features there. And then it says, This is an essential track suit built to perform and designed to look right anywhere. So let's have a look back at our custom instructions and compare. So we said we wanted it to be confident, modern, and premium in the brand voice. And I would say the tone should be clear. It was very clear, professional, approachable, and slightly aspirational. The writing style rules were short to medium length sentences, no hype exaggeration, no emojis, no buzzwords or cliches, clarity over cleverness, formatting preferences, use headings are helpful, bullet points, avoid excessive markdown. Content is from modern apparel brand selling premium track suits. So if we go back now, I'd say it's met that because it has the headings, just a few headings, bullets where they're needed. It's been very clear and to the point, very professional in its output. And all of that was from the custom instructions, and now anything that we ask it to do will match the custom instructions, making it far more helpful, far more effective. Doing what we want. Now, one way to test that is to simply ask it to do something else and see if it follows the same pattern. So I've said rewrite this for a lunch announcement. Really simple and to the point. No instructions there. Let's see what it does. So you can see, again, quite minimal and quite clean in its response. Introducing the Black slim performance track suit. Our first track suit is here, built for movement and designed for everyday wear the Black slim fit performance track brings together clean design and athletic functionality. The silhouete is modern and streamline. The fabric is lightweight, provable, and engineered move with you. There was no real need for bullets here because it's a launch announcement, but it is clean. It has put in the heading. It does sound professional, available online in limited quantities. So again, if you were to look at the two, you would see that it follows the same heading and very minimal text and professional text. So this is all in line with the custom instructions. Now it's your turn. Set the custom instructions for your own brand or role, and examples include set it as a consultant, product manager, a coach, a founder, and offer internal company documentation. Keep it short, clear about style, not task. And once this is set, you're ready for the next lesson. We're going to build on this to control the structure and the consistency even more at Scale. See you in the next lesson. 7. How To Keep Long Conversations On Track Using Context Continuity: So in this lesson, we're going to solve one of the most common frustrations that people have with Chat GPT, which is it started well, but halfway through it, forgot what you were doing and a common problem to do with the way Chat chPT works. It usually isn't a bug. It's usually a misunderstanding of how the context works within Chat GPT. Now, Chat GBT doesn't think in documents or projects by default. It thinks in conversations, and everything it knows about your task lives in the current context window. So today, you're going to learn what the context continuity is within Chat GPT, why long conversations drift, sometimes, how to structure a conversation, so Tat GPT stays focused, and a live demo of planning a track suit launch start to finish in one thread. So today, the concept we're using, the concept is context continuity. Why we use it, we use it to maintain shared understanding across a long chat. And what it solves, it solves drift where you drift into some kind of other output that you didn't want repetition, lost details, the conversation and inconsistent outputs. The important distinction here that really matters later is context continuity is the short term awareness inside the conversation. So what Chat GPT is aware of at that time, and that's different to memory, which is the long term storage across different conversations. Right now, we're just talking about the short term awareness within the current context. So let's talk about how Chat GPT understands the conversation. Chat GBT doesn't remember everything forever. Instead, it reads the entire conversation so far and it uses that as a working context. New messages are interpreted based on what came before. So think of it like a whiteboard. Everything that is written stays visible. But if you erase or overwrite things, early ideas lose importance. So if your conversation changes the direction suddenly or your conversation introduces new goals without restating the old goal, or it mixes unrelated tasks, then the important thing is Chat GPT will follow you right off track. So if you go off track, it will go off track, it will lose the context. So you're planning a new tracksuit launch. And what you want to go over is the product concept, the audience, the messaging, the assets, and the launch steps. And you want Chat GPT to stay aligned from start to finish. So to anchor the conversation early, we're going to type in the current prompt. So here we're saying that we're playing the launch of a new premium black slim fit track suit. This conversation is for planning the product messaging and launch steps. Please keep all responses aligned to this go unless I say otherwise. Knowledge and the reason we're doing is to keep Chat GPT initially really focused on exactly what we're doing. If you don't do things like this, that's when you tend to drift off top it. So you really need to essentially give it controls, give it the mindset that you want, tell it what to stay focused on at various points in the conversation. So this will get us to stay on track. So let's run that. So you can see that Chat CPT comes back, acknowledges it, and says it's going to keep all responses focused on planning, the product messaging, and the launch of the premium Black slim fit tracksuit. Awesome. The next step is to build on the same thread. We don't want to go off track, so we're gradually going to build on what we want layer by layer. So the next prompt is start by defining the target audience and the core product positioning. So there you go in the same tone that we gave it in the custom instructions. We've just got headers and bullet points kept really simple. It's given us our target audience here, primary customer, and the bullets associated with that customer. So use cases, and then the core product positioning with a positioning statement and the pillars here of versatile performance, modern fit, minimal design, premium feel. This positions at track suit as an essential, not a trend. So in the same style that we told it to in custom instruction, we got the output now. We're going to ask it some more questions to do some more things, but focusing on specific things that we want to get feedback on. So here we've said, based on the audience and positioning you just defined, draft a short product description. The important part here is based on the audience and positioning you just defined. This keeps it focused on what's above. And this is how people sometimes go off track with Chat GPT. If we were to say draft a short product description, it could use anything up here. Or it could use anything previous to this conversation. And what we want to do is keep it really focused. So we're telling it based on the audience and positioning that you define. So that keeps it targeted to what we can see here. Let's run that. We can see here said a black slim fit track suit, premium track suit, modern active lifestyles, slim athletic fit, clean silhouette. Breathable lightweight fabric, all day comfort, et cetera. And if we were to look up here, we can see it's used certain elements such as the premium field, the modern fit. It's just created like a product description, but it's targeted just the facts that we put here. The next thing we're going to do is we're going to get some image ideas for this kind of tracksuit, just give us some ideas before we create any images. But the important thing is how we say this. So have a look at how we say this to chat GPT. Again, we're referring back by saying, keeping the same positioning audience, describe three image concepts for this tracksuit. Again, if we didn't say this, we may get something that doesn't necessarily relate to what we said before. So I'm going to demonstrate that. First of all, let's run it this way, and then I'm going to show you a different prompt, and we'll see the difference. So nice clean way, as we told it to in our custom instructions is given us three concepts, one for urban movement, one for performance, detail, and one for being travel ready. And it's given us the right age ranges here, for example, 25 to 35, which, if we refer back, are within our age bracket, clean, modern using some of the words that we see here. Minimal. Again, referring back, we can always find that it refers to what we've said already. Now let's change the prompt and make it less focused than what was done before. So here we said, describe some image ideas for a track suit. Let's run that. So if you look through this, the first thing see is that it's a lot less targeted. It's a lot less targeted to what we said earlier. It's a lot less targeted to the target audience and the core product positioning. So what we've got back here is a model walking through a modern city street mid stride This focuses on natural movement and how the track suit fits while in motion, clean background, neutral tones. So if you read through these without having to through all, it refers less to some of the elements you could find in the target audience and the core product propositioning, whereas when we ran this prompt, it spoke specifically to things like the age group and the minimal nature of it. So the fact is that depending on the prompt that you put in, you can make sure that the feedback really relates to a specific point or a specific target audience, a specific proposition, or something that you want Chat GPT to focus on. By being less concrete about the fact that you want it to focus on those things, you might get back feedback that answers your question, but doesn't answer it necessarily in the way you want. So what do you do if the conversation starts to drift or you don't get back exactly the output that you want? So this is when you can reset and resent a Tat GPT back to what you want. Here's a prompt that will do that. So I've said here, quick reset, we are still planning the launch of a premium black slim fit tracksuit. Target audience and positioning are unchanged. Continue from that context. Let's run that and see what happens. And now, if we go back and say, describe some image ideas for a track suit based on the above, so now it's targeted to what we've told it for the product launch. So now if we take the first one with the modern city, it's a model walking through a clean urban environment, architecture, minimal distractions, the focus is on the silhouette movement, not attitude. A purpose communicates everyday versatility and style beyond the gym. Now, this purpose has been added, as you can see, if we go back, we can see from the target audience here that the buying mindset is one of versatility and one of the use cases is going to the gym. It's pulled directly from the target audience, a couple of points to make sure that if we go back down, these image ideas relate exactly to our target audience, and that's the beauty of saying things that get us more focused on exactly what we're trying to do. When we prompt chat GPT to focus on specific things, we get a more specific output. So the simple rules here for long conversations, memorize these. They'll save you hours. So state the goal early, build step by step in one thread, reference early outputs explicitly. Don't assume Chat GPT knows what matters. You have to tell it, and then anchor when needed as we did there. So now it's your turn. What I want you to do is pick a real task in your career or business. It could be something to do with product planning, document drafting, strategy design, research synthesis, creating research, and run it entirely in one conversation. Start with a clear goal. A clear scope, a simple anchoring message that really focuses Chat GPT. Then build step by step and always referencing what came before. And that way, you'll find that your conversations with Chat GPT are much clearer to achieve your goal and are always targeted around what you're talking about. You'll get less drift off topic. I wish you success with that, and I'll see you in the next lesson. 8. How To Remember Key Details Using Memory: Okay, so today, the question is, how do I stop repeating the same information every time I open ChatGPT? And that's where memory comes in. Memory allows ChatGPT to remember important details about you and your work across different conversations, not just inside one chat, but over time. So the feature that we're concentrating on is memory. Why we use it is to persist important details across chats, different chats. And what it solves is it does having to repeat brand info goals and preferences over and over again. The thing to remember and the critical distinction is that context is short term and exists inside one conversation. Memory is long term and across many conversations. It exists across many conversations. They both work together, but they're not the same thing. How long does memory last? Well, memory persists across chats with no fixed expiry. Save memories remain until you delete them or turn off your memory, and the only key reusable details are stored, not full conversations. You can review and manage memory at any time, and memory behavior may evolve as ChatGPT updates in the future. So this is how it behaves at the moment. So when we enable memory, then ChatGPT can save important facts about you or your work, reuse them automatically in future conversations, and reference them without you restating them every time. So in order to work with memory and adjust the settings on that, go down here to the profile image, personalization. And then scroll down to memory. What you can do here is you can first of all, reference save memories. So this lets ChatGPT save and use memories when responding. If you don't won't persist the memory long term. The other thing you can do is if you click this manage button next to memory, you'll see in here under save memories exactly what ChatGPT remembers. Now, as you can see, even though we've done so much typing and we've done a lot in one chat, memory is disabled, so ChatGPT won't use or save memories, meaning all of the stuff that we've put in here is in context within the chat, short term but it isn't in memory long term. And also, if we go between different chats, it won't remember what we've done. It may remember some pieces of it or some reference to it, but it won't remember the specifics. So let's test that right now, and we'll see that memory is disabled when we come back here. We'll also see that it doesn't actually remember in between chats. So if we come back to ChatGPT and we start a new chat, and here we just put in this prompt. And I'm saying, help me write a product description for a new track suit. It's not referring to anything that we did previously. We did say new track suit, but it does have some context about what we're doing in the other chat. It could have referred to the here it says, Here's a clean premium product description you can use or adapt. It's written to feel modern, considered and confident without overstatement. So now let's repeat the exercise after we have enabled memory. So we'll go down here, go to personalization, we'll scroll down to memory, and we'll put reference saved memories. And you can see down here, as well as saying reference saved memories now there's another option for reference chat history. And if we zoom in, we can see. It says, Let ChatGPT reference recent conversations when responding. So now, that means that if we're in one conversation, we can reference another one. So let's go back to Chat main window and let's start a new chat and say, write me a product description. And you can see here straightaway the first thing it's done is say black slim fit track suit, which is exactly the track suit that we named in our first chat window. Meaning that it knows when we talk about a product description what we're referring to straightaway. Now, you can see if you go over here to the chat side, you can see this is in a completely different chat called product description writing, whereas the first one we did here was called Track suit Launch Guide. I'm just going to rename this slightly so we can remember it, and I'll rename this one. The original conversation we had is here and this is where we went through the process of creating our product description. There's a lot in here, but ultimately, we decided the target audience and the core product positioning a premium Black slim fit track suit in here. In our second chat, it wasn't even mentioned that it was a black slim fit track suit in here because it was doing it from scratch, not from memory. And in our third chat, you can see the first thing it mentioned was the fact that it was a black slim fit track suit, and that's after we turned on memory. So that shows you that memory is working in between chats. So it's important to decide what to put in memory and what not to put in memory. What not to put in memory is things like specific launch dates, pricing experiments, temporary campaigns, and draft ideas, things that we're not sure we want to save forever. However, we want to put in memory things that are going to be retained in between chats. And here again, I've tested it in a fourth chat. Help me write a product description for a new track suit. You can see it says, The essential Black track suit straightaway it's using details from our first track, the track suit launch guide, slim modern fit, et cetera. So it's definitely remembering things that we did in our original chat. Sometimes you're going to want to force ChatGPT to remember things because there's a difference, as I said, between context and long term memory, persistent memory. So even if you're not going between chats, you want it to remember things, and telling it to remember things is very important at times as well if you want to be sure that it has done so. So if we want something to be remembered, definitely in memory, not just between chats, but persistently, what we can do is if we paste in this prompt, we've said, remember this, I run a premium track suit brand focused on minimal modern designs. I usually use GPT for product planning, marketing, copy, and launch strategy. My preferred writing time is clear, confident, professional. If we run that, you can see up here, it actually says updated saved memory noted, I'll keep this context in mind and apply it consistently in future responses. And if we zoom in a little bit here, we can see it says runs a premium trak suit brand focused on minimal modern designs. They use ChatGPT for product planning, so it's writing it from the perspective of ChatGPT, but it's essentially saying that it remembers that we run that premium tracksuit brand. Got the manage button here, and what that will do is it will allow you to manage what's saved in memory. And there are two ways to reach it. One is to actually use this manage button. If we click that, we'll see the save memory in here. And the other way to reach it is down the bottom here. If we go to personalization, scroll down to memory and click Manage, the exact same place. And what that allows us to do is actually delete things that we don't want to remember there. So because I've already got this in my custom instructions and in various places, and I want to be sure of what's where I could always come back and delete this if I wanted to. I'll leave it there for now. In summary and this lesson, you've learned what ChatGPT memory actually does, W to control it in ChatGPT, what information is worth remembering, how to save, test and manage memories. The difference between memory and context. The key takeaway is that memory is not intelligence, it's continuity, so it persists between chats and even in one chat long term rather than short term like Context does. Use it sparingly and intentionally and review it regularly. I want you to do is add one or two meaningful memories within ChatGPT. It could be things based around your role, the type of work you do, your preferred writing style. Then start a new chat and see how ChatGPT responds differently, and you will see how it responds based on memory. Once this is set up properly, every lesson that follows is going to become faster, cleaner and more consistent because you've got memory stored. So enjoy and see you in the next lesson. 9. How To Keep Outputs Consistent Using Constraints: Okay, so in this lesson, we're going to solve a very common real world problem, which is why does ChaGBT give me good answers, but not in the same format every time? So this is where constraints come in. So constraints let you control the shape, the length, the structure, and the rules of Chat GBTs output so that results stay consistent and reusable and predictable. So in this lesson, I'm going to show you what constraints are and what they're not and the most useful types of constraints in real work. So the feature is called constraints. Why we use it to control the format, the length, and the structure of the output that we get from Chat GPT. And what it solves is inconsistent answers and having to do lots of rework because things don't come out the way that we want them to. So the important framing here is that instructions tell hat GPT, what to do, and the constraints tell Chat GPT how the output should look. So let's talk about what constraints do. So constraints allow you to specify rules such as the output length, for example, Max 150 words, the structure, for example, headings, bullet points, and tables. Ordering, for example, step by step or number exclusions, such as do not include Imogis or style limits such as no marketing language. And they work inside the prompt, not in the settings. So let's do a demo and show you what I so this demo is still working with our product copy for our new product. And the scenario is that you're creating multiple product descriptions for a range of different track suits. But you want every single description to follow the same structure. It'd be easy to scan and read through and be ready to publish without editing, how are we going to do that? So let's within Chat GPT, have a look at the product descriptions that we've created up until now. So the first one that we created, we said, help me write a product description for a new tracksuit. You can see it's got a heading and it's got a paragraphs, some key details here, and then another heading, some key details in bullet points and a line there. The second one is similar in the layout. And the third one it starts with black slim fit tracksuit. It's got a couple of paragraphs here and bullet points, no bullets underneath. So you can see that this one has got key features and why it works. This one's just got key features. This one's got key details on why it works, and then the heading at the top. And they're all very slightly different. So what we're going to do now is we're going to put in some constraints that mean that there's going to be more predictability about the output. So I'll said here, write a product description for premium track suit using the following constraints. Exactly three short paragraphs. Each paragraph, no more than two sentences. Use a confident professional tone. Do not use emojis, do not use marketing hype or exaggerated claims. Let's run that. So you can see here it's written this out exactly as we've said. The first thing to notice is if we zoom in a little bit. So we can see there are no bold headings or anything here. It's literally just three paragraphs exactly what we said it would be. No more than two sentences. So full stop there, full stop there, two sentences. Same here. Confident professional tone. This premium track suit is designed with a clean silhouette and a modern slim fit. I offers a refined look that works beyond casual settings and no emojis and no marketing hype. I just gets to the point. By doing this, this means that every time we create product descriptions, they can come out in exactly the same format, saving us time or saving us worried about worrying about having different formats with different types of headings. So what we can do is we can go even further and we can specify what each of the paragraphs is actually referring to. So if we paste in our prompt. So now we're saying write a product description for a premium tracksuit using this structure, but paragraph one must be what the product is for and who it's for. Paragraph two is the key features in the material. Paragraph three is how it fits into everyday use. And we've got the constraints here, two sentences per paragraph, clear professional tone, just like we said before, and nomogs or buzzword. So this has remained the same. But the key thing is we're saying what needs to be in each paragraph now. Let's run that. So now, as you can see, paragraph one, what the product is and who it's for, for individuals who value clean design and considered details. Paragraph two, key features and materials made from a smooth, durable fabric, it balances structure with comfort. Paragraph three, how it fits into everyday use. Easy to wear throughout the day, it transitions seamlessly between travel, et cetera. So that constraint gives us more control over what's in each paragraph, and this can be used for anything that you want to use it for going forward. It doesn't have to be product descriptions. Now, you may be thinking, well, we already have custom instructions set up if you set those up and you may be wondering when you would need constraints. So for custom instructions, we use these when we want to default tone, a role or preferences that apply to everything. Was constraints are per task structure and rules. In other words, it depends on the task we're doing. Sometimes we want to slightly change the constraints or slightly change how we want to do things based on the task we're currently doing. In practice, custom instructions stay the same for every single conversation and for everything that we do in hat GPT. Constraints change depending on the task that we're doing. So now it's your turn. What I want you to do is take a task that you repeat often, for example, product descriptions, internal updates, research summaries, meeting notes, and rewrite your prompt and add three to five clear constraints around the structure, the length, the tone, what you want to exclude from it. And that way, you'll have a way to make sure that you've got consistency throughout all of the output that you get from Chat GPT. See you in the next lesson. 10. How To Turn Notes And Files Into Finished Documents Using File Uploads: Okay, so in this lesson, we're going to talk about how to turn notes and files into finished documents using file uploads. In real work, a lot of work begins with rough notes, draft ideas, supplier documents, spreadsheets and messy or incomplete information. So today, our goal is to turn real world notes or files into clear finished documents using chat GPT. So the feature is called we're calling it file uploads, the file uploads feature. We're using it to read and work with real documents, and the problem it solves is turning raw inputs or incomplete input into usable outputs. So when you upload a file, Chat GPT can read the contents of the file, understand the structure, the headings, rolls, lists, combine file content with your instructions, and produce new documents based on the input. The important limits to understand is Chat GPT isn't modifying the original file. It works on a copy of that content, and you control what the final output should look like. So in this demo, we're going to use two simple files. We're going to use the tracksuit launch notes, which is going to be rough and unstructured, and we're going to use the supplier options file, which is a bit more structured. So let's open those two files up. So here we have the tracksuit launch notes. It says, our goal is to launch our premium black and gold tracksuit line, target audience 25 to 40, urban professionals, brand values, minimal quality confidence, key features, slim fit, breavable fabric, durable stitching, and the channels is going to be a website launch with an email list, and it's going to promote on Instagram. So it's really sparse. There's not much long English language words here to the level of bullet points. So that's the track suit launch notes, textFile. And here's our supplier options file. So here we've got some names of some fake suppliers here, and it's got the supplier, the material used, the cost per unit, and the minimum order quantity is what this is. So let's take supplier Alpha textiles. The material they're using is recycled polyester. It costs 18 something. Let's assume it's dollar per unit. And the minimum order quantity is 300, urban fabrics, cotton blend $22 per unit and at least 200 units is the minimum order quantity that we're ordering, prime ware limited performance knit material, and it's $25 per unit, and we want 150 units as the minimum order quantity. So this is the information a little bit more structured here as opposed to the text file. So what we want to do here is we want to create a product launch brief from the rough notes and the supplier data. And a product launch brief is a lot more structured. It's a document that we want visually to be able to comfortably read, not just in very sparse terms or unclear terms. We want it to be really clear and comfortable to read. The first thing we're going to do is load both files, and here they are. These are the files that we need, so we'll open those, and then we've got to tell Chat GPT what we want to do with them. So if we paste in our prompt. So we've said using the uploaded notes and supplier file, create a clear product launch document with the following sections, product overview, target audience, brand positioning, key product features, supplier comparison summary, and recommended next steps. This is a professional document that we're creating from these two smaller, more vague documents. So let's run that. We've also said to keep a fasional tone and be concise. Great. So this is our document. So now, it says, premium track suit launch document. It's given a product overview. The product is a premium black and gold track suit designed as a core statement piece for the brand. It combines a modern slim fit with restrained detailing to deliver a refined everyday set. And I won't go into this, but it's basically giving us a target audience, brand positioning, key product features, some player comparison summary, and recommended next step. But the main thing is it's all in a document that we can actually download. So if we click Download here, you can download it as a PDF, a Word Doc or Markdown. So it gives us the output in various formats that we want. And if we go back to the original documents that we uploaded, you can see we literally had supplier options, and we had the gold, target audience, brand values, key features, and channels, nothing anywhere near as readable as what we got back. Which is this. Now, because all the information is within Chat GPT, what we can do now is create as many documents and as many different types of documents as we want. So now it's your turn. Take a real file you already have like meeting notes, a draft document, a spreadsheet, a supplier list, upload its Chat GPT, and ask Chat GPT to clean it up, summarize it, and turn it into a finished document. Once you're comfortable with this, you'll stop starting your work from blank pages altogether, and you can start them from documents you already have. See you in the next lesson. 11. How To Check Your Output Using Verification: Okay, so in this lesson, we're going to solve a critical problem, which is, how do I know if Chat GPT's output is actually correct? This is where verification comes in. So verification is the practice of checking, challenging and validating Chat GPT's responses before you trust them. So with verification, we use it to validate accuracy and claims. We use it to solve over confidence and hallucinations. So over confidence being that Chat GBT thinks it knows what it's talking about when it doesn't or hallucinations, meaning that it's actually making up information just to try and be correct and give us an answer. Verification isn't a setting or anything. It's a technique that we're going to apply through our prompts. Why is verification necessary? So it's necessary because Chat GPT is very good at writing confidently. It's very good at filling in the gaps, and it's very good at sounding persuasive. But it's not guaranteed to be factually correct. It's not guaranteed to cite real evidence, and it's not guaranteed to know when something is actually unknown, and it should tell us when something's unknown. So that's why we apply verification, and it's our responsibility to make sure that we verify information to make sure that it's correct. So in this demo, what we're going to do is we're going to take the product description that we've already created, and it's going to be deliberately flawed. And it's going to include overstated claims, unverified statistics, and vague authority references. And that's exactly the kind of thing that Chat GPT often produces. So let's take a look at it. We've got a draft product description, and it says, Our premium track suit is made from the highest quality materials and is guaranteed to be the best track suit on the market. It improves athletic performance by 40% and is trusted by professional athletes worldwide. The fabric is eco friendly and sustainably sourced. So that's the claim that we're making. Let's go back to hat GPT so we can verify it. So the first thing we're going to do is upload the description. Plus button here, find the file. There's our product description, put in a prompt to verify what's going on in there. So it says, review the uploaded product description and do the following. Identify any claims that require evidence, flag statements that may be exaggerated or unverifiable, highlight anything that could be misleading. Suggest safer alternative wording where needed. Do not rewrite the full document yet. So that's very clear what we need to do. Let's run that. Here we can see Chuck GPT said Below is a structured review of the uploaded product description. Risk focused on risk, clarity and compliance. It's picked out some claims that require evidence like it's made from the highest quality material. This is a comparative quality claim. It would require clear criteria or third party validation. The fact that it said it's guaranteed to be the best track suit on the market, is an absolute claim that would require substantiated market wide proof, and it carries on improve athletic performance by 40%, trusted by professionals, et cetera. Then it's gone on to speak about potentially exaggerated or unverifiable statements. Some of the same things. It's gone on to talk about potentially misleading elements. The performance improvement claim could mislead customers into expecting measurable athletic gains. Sustainability language, guaranteed, the fact that we said guaranteed and then it suggested safe alternative wording. So it's indicating that we could reword this if we wanted to by replacing absolute quality claims with things like made from carefully selected premium materials, things like that. So at the end here, it says, The draft contains multiple high risk claims from a legal and brand credibility perspective, addressing them early will reduce exposure while maintaining a premium confident tone. So this is helpful because it shows where we could have made too many assumptions, made too many claims that could land us in trouble later. And the text is actually something that could easily be generated by hat GPT if we are not checking things over carefully, and this shows how we could check it over carefully and then refine it later. So now let's refine it based on the feedback we've been given. So let's have a look at this prompt. We've said rewrite the product description using only claims that can reasonably be supported, remove or soften anything that cannot be verified, keep the term professional and factual. Let's run it. So now if we zoom in a little bit, you're saying that this premium track suit is designed with a clean, modern silhouette and focus on everyday wearability. It reflects a considered approach to fit material selection and finish. The fabric is selected for comfort, durability, and a smooth feel against the skin. Construction details prioritize consistency and long term shape retention through regular use, minimal branding and restrained detail and keep the design versatile. The track suit is intended to perform reliably across daily routines from travel to downtime. So it's talking about the intention without actually making any guarantees in me, that's probably a bit too soft now, but it just shows what we could do if we wanted to to soften anything that cannot be verified. And we can always make it a bit stronger in places if we want to. The idea of this is to understand when there are some things we can do commonly to verify claims in our text whenever we want to. So you can apply verification anywhere by asking things like, parts of this answer might be incorrect? What assumptions are being made here? What would need evidence to support this? Where could this be misleading? And verification is a kind of habit that you want to build into everything you're doing. You don't want to make it a one off step that you do once. You want to check when there's a need to verify anything. So the reason this is a good step, and it matters for any work that you're doing is if you don't apply verification, it can lead to incorrect decisions, legal risks, loss of trust, reputational damage, whether you're a business or just working within a company using Chat GPT for anything you're doing. You really want to verify the output at the right moment, and you can decide when the right moment is, but I would suggest you do it regularly. And instead of something that just sounds good, you'll end up with something you can safely use, whether you're using it at work or to run your own business. Now it's your turn. What I want you to do is take something GPT created for you recently, could be an email, a document, a recommendation or a claim and ask hATGBT to identify the assumptions, flag any risk and suggest safe alternatives. And once you do that, you'll be using CHAT GPT as a powerful assistant that can not only generate content for you but verify that content, so it's safe to use anywhere in a workplace or in your business. See you in the next lesson. 12. How To Analyse Data And Spot Patterns Using Data Analysis: So in this lesson, we're answering the question. How do I use hat GPT to understand data without being a data analyst? At the moment, you'll probably see that there are a number of job roles around for a data analyst, and data analysis is about working directly with data and pulling out relevant information. So in Chat GPT, using data analysis, it allows you to work directly with spreadsheets and tables and other information, ask questions in plain English, spot trends, patterns, and anomalies, and turn raw numbers into insights. So this feature of data analysis built into Chat GPT, why we use it, we use it to explore and interpret structured data. It solves the problem of making sense of numbers quickly. And Chat GPT knows how to apply data analysis effectively as soon as you give it some information that consists of data like numbers or a structured file like a CSV or a spreadsheet. So when data analysis is in progress, hat GPT can do things like read tables and spreadsheets, group and compare data, calculate totals, averages, and changes, identify trends over time, and explain patterns in plain language. In other words, you can interrogate the data that you've given Chat GPT. Here, what we're going to do is we're going to use a simple sales dataset. So we're going to use some sales data in this CSV. And the file includes product names, monthly sales, units sold and revenue, and it's intentionally small and realistic. So we can just show an example here. So let's download that. So here you can see the tracksuit sales data. And we've got the product. It's the same product each and every time, and we're showing for different months the amount of units sold and the revenue. So the BlackGold tracksuit in month January, it sold 120 units, and the revenue was $8,400, February 150 units, $10,500, March, 210 units, $14,700, et cetera. So it's literally just going through each month and the number of units. So on with the demo, in any case, we have is a scenario where we want to understand which products are performing best, whether sales are improving over time, and where to focus next in terms of increasing sales. So the first thing we're going to do is we're going to upload the document, and then we're going to prompt Chat GPT to help us to do these things. So let's upload our document, and we're going to paste in a prompt I created earlier. Let's have a read of that. So it says, based on the uploaded data, which product is selling the most units overall? How are sales changing month to month? Are there any clear trends? It'll be interesting to see how it handles this document. So let's go. So let's see what came back from Chat TPT. What it said here is that based on the uploaded sales data, here's a clear summary of performance and trends. Top selling product is the gold tracksuit, is the strongest performer. It has sold 480 units compared to 305 units for the Black silver tracksuit. So let's see what happened. It was a black gold track suit and a black silver track suit, and it's comparing them different for the same months for two different track suits. We can see if we total these up, but the sum is 480 for the black gold, and the sum here is 305 for the black silver tracksuit. So let's head back to Chat GPT and keep reviewing. So Chat GPT says that this indicates clearer product market alignment for the black gold color way. In other words, it's a longer way of just saying it sells better. So the black gold track suit is selling better than the Black silver one. Month to month sales changes. Both products show growth over time, with March being the strongest month. Black gold track suit, January to February increased 120-150 units, February to March, strong increase to 210 units and it tells us the same for the Black silver track suit. Clear trends. Demand is accelerating rather than flattening particularly in March. The BlackGold track suit consistently outsells the alternative colorway each month. Sales momentum suggests growing interest rather than one off purchases. Overall, the data supports prioritizing the black gold track suit, et cetera. So this is just a really basic simple analysis of the numbers. So now we can go a little bit further. Let's type in a new prompt. So here we're saying, compare the growth rate of each product, which one is growing faster and why. It'll be interesting to see what Chat GBT says here. So let's run that. So here, it's showing us the comparison. It's reiterating the amount of units. So the total growth for the Black gold track suit has been plus 90 units between January and March. And the growth rate is around 75% or 75% over the period. Black silver has grown 50, 62.5% over the period, which is growing faster. The black gold track suit is growing faster in both absolute and percentage terms and also month to month, particularly from February to March. Now, the likely reasons, the black and gold colorway aligns more closely with the brand's premium positioning, higher early sales created momentum and visibility, clearer customer preference compared to alternative color options. Overall, black gold is not only selling more but doing so at a faster rate, reinforcing it as the stronger product to prioritize. What it's basically saying is we said that we were going to sell black and gold, and that's the way we positioned it. Obviously, the customers seem to have gone with what we positioned it as in the first place. It's saying likely reasons, it really doesn't know, and so it's just making some assumptions, really, but at least we know that these are assumptions, let's paste in another prompt. And if we zoom in, let's have a look at this. It says, summarize the key insights from this data and suggest two practical next steps. Go. So we've got back some insights. I don't think it's too surprising, but this is just to demonstrate the way Chat GPT will respond to things like this. So it says essentially it summarizes that the black gold tracksuit is the stronger performer. Both products are trending upwards, customer preference appears clearer over time, suggesting product market fit is stronger for the black gold. We know all that from the above. And the next steps it's saying is just essentially prioritize the black gold tracksuit. But it's also saying test and refine the secondary color way. So use targeted content or limited drops to assess whether demand for black silver can be improved. You still want to apply everything that we learn in our other lessons, such as file uploads, verifying checking conclusions, applying constraints. And also, once you've mastered using everything that Chat GPT has to analyze data, you can use it for research planning and decision making. So now it's your turn. What I want you to do is take any spreadsheet that you already have. It could be sales data, expenses, timelines, performance metrics, upload it into Chat GPT, and look for what stands out, what's changing in the data, and what you should look at next. Ask Chat GPT that and allow it to give you some suggestions. Once you get comfortable with this in Chat GPT, any data you have no matter how complex will stop being intimidating because you know that you can interrogate the data and ask the right questions to get the right answers so that it's really useful for you. So that's that. See you in the next lesson. 13. How To Research Topics Quickly Using Web Search: Okay, so up till now, you've used Chat GPT to think plan, generate based on its built in knowledge and your instructions and asking questions. But in this lesson, what we're going to do is we're going to ask hat GPT to actively search the live web for information. And that's essential when you need up to date facts, current trends, recent announcements, real world validation before making decisions. So the feature that we're using is called web search. While we use it, we use it to pull live current information online, and what it solves, it solves the problem of outdated data assumptions or guesswork within hat GPT. Let's talk about the difference between web search versus normal chat. With normal chat, it uses Chat GPTs internal training and memory. This is best for reasoning, drafting, planning, and synthesis. It might be outdated for recent events or trends. If you can imagine hat GPT is trained in a whole bunch of data, including the data that we give it, anything outside of that, it's not actually going to see even if it has gone to the web to get that information, it's not going back there necessarily every time we ask it a question. With websearch, it actively searches live websites, and that's best for current trends, announcements and comparisons to real up to date data, but it's verifiable and it's based on a particular source. The rule of thumb here is if you want chat GPT to use the information it's got to think first, then use normal chat. But if you want to check facts or trends, especially current trends, then use websearch. So let's get into a demo. What we're going to do is we're going to research tracksuit market trends using web search. So let's say we're planning a new tracksuit line as we are, and we want to know what styles are currently trending in the marketplace. We want to know what materials are popular, and we want price ranges that the competitors are currently using. This requires live market data and not assumption. So if we were to use normal chat we'll just use the information that it's already got, but we want to use the live market data, so we want to go to the web search to do that. So in order to do a web search, we'll go here to the Plus button and choose more and then go to websearch. And what we're going to do is we're going to paste in a prompt I made earlier as usual. And if we zoom in a little bit, the prompt says, What are the current trends in premium track suits for 2025, 2026? We're going to focus on styles, materials, and price ranges. Let's run it. So here, Chat EPT gives us a concise summary of current trends in premium track suits of 2025 and 2026, exactly as we've asked for, focusing on the styles materials price positioning, and it's grounded in reporting and market insights. So here, if we zoom in a little bit, the styles are modern silhouettes. And well, this is all about the styles, modern silhouettes and design language. It says that the track suits are evolving beyond basic athletic wear towards fashion forward silhouette, and it's using sources here. So if you look here, you can see it's using Ali Baba. Some more information here about the fact that there's retro and streetwear influences remaining strong with vintage inspired cuts and nods to classic sportswear. It's got axio link to axio there and some fashion trend reporting highlights the return of retro track pants. If we check out these links, these referencing links, I'm going to open them in a new tab and let's have a look at those. So here we can see some examples of reports from reputable places like Alibaba. Now. If you don't know, Alibaba is where a lot of brands will go to buy clothing and any product, in fact, wholesale. Alibaba is going to have a lot of information about what all of the brands are buying, what the companies are actually buying. And in that case, I can tell very reliable source because people from all around the world, particularly America, England, as it's based in China, even China, go to Alibaba to purchase products. So it's reliable information, and what they're saying is in 2025, fashi tracksuit versus classic track suits, are the new styles really worth it? So, this was a report on that, and then it's got some of the kind of track suits here in an image. Jim to glamour. And so, essentially, what Chat GPT has done is it's referenced this report. You can see it's got things like average cost of the track suit, classic track suits, $6,220, 225 fashion track suits, 150 600. So what we can do is we can trust now that we've looked at these pages that it's based on a reputable source. And we can do the same for these other websites. We can look up, you know, the validity of somewhere like at CR, see whether these such volume trends are things that we can trust. And therefore we can use that to come to some decisions for our company, and the same with this one from who, what, where.com Seven Pant trends coming for your wardrobe in 2026. So coming back now to Chat GPT, we've got now some information that is based on live data, and we've got links to the website so we can check them out and see if we want to trust that information or not. I can definitely tell that the Alibaba information coming from a good source. And then if we look down materials, technical and performance blends, premium track suits are increasingly made with engineered textile blends, such as stretch knits and moisture management fibers, price ranges, segmented pricing tiers. Track suits span from mid range offering, 75 to $200 to higher tier, luxury segments, 200 and above. So without going through all the information, what we can see is it's poured back some information from the web, and by doing it in this websearch, we guarantee that anything we search for is from the web. In a normal at, we'd have to be very specific. But now, but this guarantees that everything we've got in here is coming back via a web search. With that, let's probe a little bit deeper. The next thing is what should we gleaned from this information. There's quite a lot here and we want to know what should be the takeaways. So let's paste in a prompt that will help us with that. I've said here, summarize this research into five clear insights I can use to guide a new tracksuit product launch. Simple as that, let's go. Then in summary, we've got five clear actionable insights from the research. One, focus on sustainable materials and transparent practices. Two, balance contemporary style with heritage cues. Three, premium quality and durability. For versatility and hybrid wearability and five collaboration and customization. Of the product. So these are things that I could go through if I'm creating a new brand and a new product and just make sure that I'm aligning with these things for the best chance of sales. So this is helpful, and it matters because from this point onwards, we know that any research will feed our decisions, and the decisions is going to feed what images we choose, what videos, what promo we do for our ads. Any mistakes can become expensive lessons for later. So the web search actually helps us to validate our assumptions, reduce risks, and then make really defensible decisions if somebody within our company or anyone at all is going to ask us why we made those decisions. We've got real research that we can use to back it up. Now it's your turn. What I want you to do is use the web search mode to research something for your career, for your company. It could be a competitor brand, it could be a trend in materials, it could be a pricing strategy or just something that you need to do. If you're going for a role at the moment, it could be researching the best companies to go for and then switch back to normal chat and then you can ask, based on this research, what would you recommend I do next or you could break it down more and ask more questions? Have a go and I'll see you in the next lesson. 14. How To Do Structured Research Using Deep Research: In this lesson, I'm going to show you how to do proper structured research instead of trying to get quick answers to your problems. What you need in this situation is something called deep research, which is a feature within Chat GPT. And that gives you a comprehensive view of the topic. It allows you to analyze multiple sources and to get clear structure instead of scattered facts. And it's something you can actually base your decisions Think of this as research you could confidently present to a team or stakeholder. So the feature is deep research, why we use it is to get structured multisource analysis, and what it solves is the problem of when we get shallow answers or fragmented research from different places, it really brings it all together. So let's talk about deep research versus web search. So when you do a web search in Chachi BT, it's fast, it's question based, and it's best for checking facts or trends. Now, deep research is actually a little bit slower, but it goes a lot deeper in terms of research. It's topic based, so we're researching a specific topic, and it's best for strategy planning and decisions. So the rule of thumb here is, if you want quick answers, then use web search. But if you want deeper strategic understanding about something, then use deep research. So in our demo, we're going to do some structured research for our tracksuit brand, and the scenario is we're planning a new premium tracksuit line as we are. And we need an answer for the following. Is there demand, what styles are dominating, and what price points work? What materials matter, and what gaps exist in the market. So this isn't a one question problem, and even though we could definitely use a web search to get back information, some quick information on this, we really need confidence, deeper confidence in this because it's going to drive what we do for our product. So that requires deep research. So the first thing to do is we need to activate deep research. And the way we do that is we go here to our plus and we click on Deep Research. And you can see here it's changed to research, which shows us this is on. So now let's paste in our prompt. And if we zoom in a bit, we're saying to Chat GPT research the premium track suit market with a focus on consumer demand trends, popular styles and fits, common materials, and sustainability factors, typical price ranges. Key brands and positioning. Summarize the findings clearly and objectively. So it's going to go and look into these things and look a lot deeper than we did with our web searches, and will actually do some of the work for us. So let's run that. So it's come back with some questions here. So that allows us to make sure to make sure that we're getting accurate responses, I asked, which geographic market should I focus on? We're going to go with Europe again? Are you interested in men's wear, women's wear or Unisex tracksuits? Is there a specific time frame you're targeting, for example, 2026? Let me know so I can provide a focused and relevant summary. Okay. So now let's respond. Okay, so we've told it that we want to focus on men's wear track suits, and for 2025-2026 trends. Let's see what it comes back with. So it takes a little bit longer, as we can see, it says starting the research because it's going to go deeper. But it says, got it. I'll research the European premium men's wear trends, men's wear track suit market, and it's saying the exact time frames that it's going to focus on, which is what we've given it. It says it's going to summarize everything clearly and objectively. So let's see what happens. As this takes a while, what we see on the right hand here is the activities panel pops up and it says, European premium menswear track suit market trends 25 to 26, and it gives you updates on what it's doing. So it's currently gathering insights on the reasons behind premium menswear tracksuit demand, considering its evolution and current trends, which is leading it to explore other factors driving interest and popularity. Says it's examining two track suit sales datasets to identify trends in the premium menswear market. Reading from track suits sales data, which is what we gave it earlier. So this shows it's employing the memory from what we've given it before. And so it's reading the file, and then it's going to use that for insights. And down here, we can always stop the research or we can update what we've told it to do here. In this panel, if you click on it, we can see there are two sources that it's using. It's using our tracksuit data from earlier we uploaded two documents here and it's using them. And it's using our track suit notes. Now they've just popped up. Everything that we've uploaded is being used has information that it's going to use for our research. So it can compare the sales data we've got with what's going on in the marketplace and give us some detailed feedback. So this is definitely very different to doing a web search. That's why I want to get across. If you can see here we're only a quarter of the way through, so we're going to wait for that to conclude, and we'll be right back. We've been waiting about a minute, and here are some more updates. Since then, it said that it's noting the need to explore current consumer demand, et cetera. Search for European men's wear track suit market trends, noting articles as well on men's tracksuit trends. So you can see that it's really searching really broadly. It's not simply doing a web search and bringing back a few pages. Analyzing market report, so I'm really happy to see that it's doing that because it means that there's a stronger value in actually doing deep research. Let's let it do its thing. Okay, so at first, it didn't generate the report. So I started asking it to summarize the report because I thought from research, I had to do that. But what happened was about an hour later, it finally popped the report up. So I'm not sure if that's a bug in chat GPT. It didn't actually tell me it wasn't going I was going to take some time. It just said it had completed. We've got the actual report, and it's exactly what we asked for. It starts off by saying this is a premium mens wear track suit report for Europe, 25 2026. And what we can see is the headings, consumer demand trends, popular styles and fits, materials and sustainability at price ranges, key brands and positioning. And then it gives some sources as well, so it tells you where it got its information from, and it's also given its sources throughout the report. So I'm not going to go all the way through this, but just to give you a flavor for what the report gives, you can see it's very detailed, and what it's saying is it split this into subsections. So, for example, athsia and comfort, it shows that demand for track suits is driven by the continuing Alesia craze. So if we zoom in a little bit here, we can see that it goes into detail about the fact that Europeans want clothes that feel like gimwar but look stylish for everyday use. It also talks about the fact that hybrid work from home lifestyles and wellness culture mean men use track suits for both exercise and casual occasions. And then it goes on to give some stats. It goes on about younger urban consumers, digital influence, about the retail growth and social media bars, sustainability and ethics. So it really is much deeper research. So I feel like it was worth doing this rather than doing a normal web search. Popular styles fits, varied silhouettes, slim, tailored. Designer minimalism. So it talks about European luxury labels and how they're reinterpreting track suits, streetwear and graphics. So you can see it's really detailed. In terms of material, it's in traditional track suits, use cotton polyester or nylon. Recycled or organic fibers. So premium men's wear track suits are increasingly made with sustainable materials. So it really is a deep dive into something like a track suit, which could be really very high level. Here's a track suit, here's how much it costs. You could give me a category, but this is a really deep dive. In terms of price, which is very important, it shows that the mid premium segment, many quality men's track suits in year at retail from roughly 100 to 250, but actually goes on to talk about much higher prices design a luxury, things like Balenciaga and Fendi costing 500 euros to 1,000 euros. So this really gives us a lot of information. It talks about marketing, positioning. Then it talks about the key brands that are out there. Nike added and Puma, street ware hype. So it's really deep. That's what I'm getting at here. We've got all the sources we can go back to, so I'm happy that it's done a real deep dive into there. So just to let you know, if it takes you a long time to run this report, it did the same for me. And in the meantime, before doing that, if I scroll you can see that I actually gave it another prompt to say I want to make a clear business decision. The decision is whether to launch a premium minimalist track suit line this year. I already have research on market demand pricing competitors and risk helped me turn this into a clear decision document. So I did that while I was waiting, and it gave me a decision brief. And the summary was to proceed with the launch, provided the product is position clearly. And so therefore, I just took this. It gave a pricing reality that the range for pricing for mid premium is 150 to 300 for upper premium, 300 to 600, and that's about enough for me. So now it's your turn. What I want you to do is use the research mode to research one of the following, either a competitor category. So no matter which business you're in, maybe you want to research the whole of a category. So is it track suits? Is it clothing at all? Is it some kind of an app you're building, and you want to research what's going on in that app category? Maybe it's a business app or app for tracking your weight. And then based on that, you can ask the question. Based on this research, what decision should I make next? So that gets us in a habit of doing some research, coming to some kind of an insight, and then making a decision. And that way, you will be able to make decisions based on the best research on the Internet, and you know that you will have done your research before you make your decision. So enjoy, and I'll see you in the next lesson. 15. How To Generate Images: So in this lesson, we're going to move from ideas to visuals. Image generation allows ChatGPT to create new images from scratch based on text. So we're going from text to image. And that's without any uploads or editing. You just describe what you want clearly enough, and ChatGPT will generate it. So the feature is image generation. Why we use it to turn text into original visuals. The problem it solves is creating early concepts or mood boards or ad imagery or any kind of imagery that you want. What makes a good image prompt? So image generation is much less forgiving than text. So a good image prompt usually has to include certain things. So the subject, what's in the image, the style, is it photorealistic? Is it studio? Is it lifestyle, Illustration? We need to be more specific because there's so many variations in an image, and there's so many ways so many things an image can show. The context as usual. So where is the image? How is it being used? And the visual constraints. So what color should it be or what colors are included? How should it fit in terms of clothing? What mood are we going for the angle, all that stuff? So bad image prompts are vague and good image prompts are descriptive but really focused on what we want in the image, really. So now let's move into our demo. So the demo is generating a track suit concept image or a number of track suit concept images. And the scenario we're in is we're planning a new track suit line as we have been, and we want lifestyle imagery. We want it to be, as we've said before, clean, premium looking, and it's got a clean premium aesthetic, neutral colors, and no logos yet. We haven't even got that far. We just want to go for a specific look. And this is exactly the kind of early stage work image generation is perfect for because we can very quickly knock up some concepts and make some decisions. Back in ChatGPT, this is normal chat, and we've just decided we don't want to use normal chat for consistency reasons. So to enter create image mode, we click on the plus sign as usual, and then we want to click on this create Image. And you can see here that it's added this image tag here just to say that everything we're doing now is going to be based around creating an image. So now we want to type in our prompt. And this shows that it's a premium lifestyle photo of a modern track suit worn by a model, and I'm going to be more specific, a male model, and it should be minimalist in design with neutral colors and slim fit in line with our brand, clean studio lighting, white background, no logos, no texts, high end fashion photography style. Awesome. So let's see what ChatGPT comes up with. So we've sent the prompt as usual. This will take a little while while it generates the image because it's a lot more processor intensive, so we'll come back when we're done. There you go. That took about a minute or maybe even less. So what we've got is this image, nice image there. And let's compare that to what we said. Let's first have a look at it. So there it is. Looks pretty good. So if we go back, so we said, minimalist design that's about as minimal as you can get. Neutral colors, it is that slim fit, clean studio lighting, white background, no logos, no texts, high end fashion photography style. I think it's checked all of those boxes. I think the image of the man's face could be more realistic, but it's definitely enough for the level we're at now. So now we're coming back to image mode, and we're going to create a variation of this by saying a premium black track suit with gold accents, worn casually outdoors, urban setting, soft natural light, modern confident athletic look, no logos, no text. Let's go. So there you go. The image is created. It looks fine. I think it looks a bit more realistic this time. The quality is a little bit better there, and it's exactly what we said. It's sleep. I think it actually could be more premium in terms of material and things like that. But I accidentally actually pasted the prompt into another window, and I think I got an image that I prefer, so let's have a look at that. So this is the one that was generated. It looks a little bit like added *** with only two stripes. But I think I prefer the look of this one is something that I think I could work with to make it look a little bit more premium. But, you know, we can do that a little bit later. The main thing is, we've got a thick that looks right and some colors that I think look right and work well with this kind of model, which is the kind of model that we might use. So some common beginner mistakes that we need to avoid is generating images in the normal chat without intent. Sometimes you can generate images in the normal chat. That's fine, but you need to know why you're doing it and have the correct intent. Sometimes you overload prompts with contradictions. So you might say you want it dark but in the daytime in the same sentence, and then ChatGPT will have to make a decision unless you say that you actually want the track suit to be dark while the lighting is bright and really specific about that. Um, you might be trying to fix images inside the same prompt. It's better to actually just generate the concept and then improve it later, like we've done. I haven't gone into any improvements. I've just looked at a couple of colors and different settings. And then expecting production ready visuals, there's usually a process to create production ready visuals. You wouldn't necessarily do it all in ChatGPT or sometimes not at all in ChatGPT. You might take it into another piece of software or you might just take it into another AI model to do that. So we generate our image concept first, and it's not necessarily the final output. So now it's your turn. Using the create image mode generate two different image concepts for the same product, one in a studio style, one in a lifestyle style. Don't edit them yet, compare the mood, the visual clarity, and that's exactly how professionals use image generation in real workflows, and we'll do more in later lessons. 16. How To Improve Images Using Image Editing: This lesson, we're going to solve the problem where we say to ourselves, I like the image that ChatGPT generated, but it's not quite right. We need to make some edits. Maybe the color is off, maybe the logo placement isn't right. The style doesn't quite match the brand. Something needs changing in there, but this is where image editing comes in. So in this lesson, you're going to learn how to edit existing images instead of starting again. So the feature we're using is image editing. Why we use it to modify parts of an existing image. What it solves is fixing visuals without starting over. And the important distinction is that image generation is creating an image from nothing or just from some text. Image editing is improving what already exists. So in our demo, the scenario is improving a track suit image without regenerating it. The scenario is that we've generated a track suit concept image, but the color isn't quite brand correct and the logo doesn't actually exist at the moment. So the overall style needs refining, and we're going to fix that step by step. So the first thing we're going to do, we've come back into ChatGPT, and this is what it's given us. So quality is pretty good, but the image itself, the quality of the track suit itself isn't what I'd want, especially for a more premium style track suit. I did actually by accident, create a different image, and we can flick to that. Here it is. And this one looks actually a lot better apart from the fact that it's got the two stripes there, which look a little bit like Addidas. So I'm going to download this one and get to work on this. There it is. So the first step is we want to grab the image that we want. There it is uploaded, and now we'll type in our prompt to edit the image. So we've put in the following prompt, Edit this image. Change your tratsu color to a deep black with subtle gold accents. Remove the two gold stripes on the sleeves. Keep the overall shape and pose the same. Make the design look premium and minimal. Do not add text or change the background. So looking at this, I just felt that the black wasn't quite deep enough. I didn't really look premium to me. And so I said exactly what I wanted here. So let's run that and see what happens. Now, because it's an image, this obviously takes a little while to run, so we'll come back when it's done. Here we are, and now we have a updated image. So I can see that's a much nicer, darker black there, and that makes a huge difference. If we look at this one, that was the before. And here, this is the after. So I'm much happier with that. So that's a good first step, but we want to refine it slightly more. Let's add a new prompt. So this is the prompt. Let's have a look at what I've written here. So I've said, keep everything the same, but add a passion sports logo. Give the material almost velvet feel. The gold and the material should have a luxury look and feel. Let's make this a bit more specific. So now I've updated it to say the gold zips and the black material should have a luxury look and feel. I've deliberately not specified what Passion sports logo should look like, so let's see what it comes up with. There you go. So it's updated yet again. So now we've edited our image. Let's talk about what image editing is best at. So image editing works best for color changes, style refinement, logo replacement, removing unwanted elements, adjusting realism or lighting, and making images more on brand. It's perfect for product concepts, marketing visuals, even add mockups and website imagery draft. Now it's your turn. What I want you to do is upload an image and try changing colors only, adjusting the style without changing structure, refining the realism or the tone of the image. Use one instruction at a time and keep refining until you get to your goal. That's exactly how professionals iterate. So that's the end of this lesson. See you in the next one. 17. How To Check / Analyse Images Using Vision: So in this lesson, we're going to look into how to check images using vision. We're solving a different problem now, we're solving a problem of how do I check whether an image is actually correct before I use it. So we know that images can look good at a glance, but at the same time still be off brand, inconsistent or misleading. This lesson is going to show you how to use ChatGPT vision capability to review images. So the feature we're using is called image vision, why we use it to understand and evaluate images. And the problem it solves is catching visual mistakes nice and early using AI. The distinction is that image editing is about changing images, but vision is about analyzing images, and today's lesson is about checking, not modifying. So what we're going to do is we're going to upload an image chat GPT. It's going to have one or more of the following problems, maybe an incorrect brand name, poor contrast text, an unexplained design element, potential brand inconsistencies. It's designed to look almost right like many real world assets. And then we're going to get ChatGPT to highlight where the issues could be. First of all, let's upload this image. This one here. And now let's put in our prompt. So the prompt that we've put in is describe what you see in this image, focus on branding, texts, colors, and notable elements. Let's see what it comes up with. So here we can see clearly it says the image shows a male model wearing a minimal monochrome track suit in light neutral tone, close to Bige. Colors and materials, the track suit is single, consistent color across both the zip, hoodie and joggers. The fabric appears smooth and matt with soft structured finish rather than a shiny or technical carries on looking through no visible branding logo or text on the garment. The design relies entirely on cut, color, and fit rather than graphical elements. So it's clearly seen what's on there. The hoodie features a full length zip. It clearly sees what's going on there. Let's upload another image. So let's upload another one of the images. This one will do, and let's repeat the exact same prompt. Run. So here we can see some differences. It's noticed that it's deep black. It's also noticed that the logo is styled stylized letter B with the word athletics printed below it in uppercase. So now with that, let's ask ChatGPT to check both images for brand in consistencies. Here's our prompt. So here we said check both images for branding consistencies, readability issues, elements that may not belong, anything that could confuse a customer. Let's go. So here, this is what it's given us back. So it said, the brand presence differs sharply. Image one, the light track suit has no visible branding at all. Image two, the black track suit, prominent gold logo and athletics text. This creates confusion around whether the brand is intentionally logo free and minimalist or a branded performance lifestyle label. And the risk is that customers may not understand what level of branding to expect from the range. It also says readability issues, the gold logo and athletics text in Image two are readable at mid range, but may lose clarity at smaller thumbnail sizes due to the thin lettering, the metallic color against dark fabric. So it's already pointing out things that we may find later due to the sizes of our images for promotion. So if we skip through this and skip down here, we can see essentially the key takeaway here is the main issue is the lack of a unified brand language across both images to reduce confusion, decide whether branding is absent, subtle or featured. A line of material feels smooth versus plush with the intended use. Ensure naming and logo style support lifestyle positioning, not mixed categories. So visually, both images are strong on their own. Together, they need tighter alignment to tell one clear brand story. So, in other words, the two track suits are two different. So let's talk about what vision is best. Vision works well for spotting incorrect logos or texts, checking contrast and readability, identifying visual clutter, detecting inconsistencies across assets, and reviewing generated or edited images is especially useful before publishing or sharing an image. What vision isn't so good at is a design approval a replacement for human judgment or a guarantee of compliance or legality. It's more like a second pair of eyes for something quick and consistent and unemotional, but a human in the loop should always be able to give that final approval rather than just using vision. Now it's your turn. What I want you to do is upload any image you already have. It could be a product mockup, a marketing visual, a generated concept of some sort. And ask ChatGPT to tell you what looks off, what could confuse someone, depending on who you're selling it to or what you're using it for. What doesn't match the intended brand? Once you get in the habit of doing this, your visual mistakes will stop slipping through your process. So hope you enjoyed that and see you in the next lesson. 18. How To Organise Work Using Projects: Okay, so up until now, we've been working with ChatGPT one conversation at a time, and that's fine for quick task, but it breaks down fast when work spans days or weeks, or there are multiple different conversations within one conversation area relating to the same goal. You're juggling research, images, decisions, drafts, all kinds of things all in the same conversation. So this lesson is going to introduce a feature that lets you group related work together. And the idea is that ChatGPT understands that all of this belongs to the same thing, even though we can separate it out and organize it in a way that's helpful to us. So the feature that we're using is called projects. Why we use it to group related chats and files together. And what it solves is when we have scattered work, lost context, or duplicated effort. Basically, think of projects as a folder of related thinking where we're grouping things that we are thinking about together. But we can still organize them. So let's talk about what a project is and is not. A project groups multiple chats under one goal. It holds files, notes and contexts related to that goal. It helps GPT to understand ongoing work without repeating the a project is not a replacement for memory. So just because things are grouped together, it doesn't mean that they're going to persist longer in terms of in the same way that memory would. It's not a custom GPT, which will come to later. It's not an automated workflow, which means that you're tying together different apps or different nodes or different tools to work together. It's literally conversations, and it's not a long term personal profile, so it doesn't make it long term just because it's in a project. So the scenario we're going to cover is continuing with our theme of organizing a track suit brand launch. The scenario is that we're launching a new track suit line, and over time, we're going to need to research trends, generate images, and draft product descriptions and make launch decisions. And if we keep doing that either in one big long chat or in random chats, the things can start to fragment and it's harder to keep track of things, especially if they're related. So a project fix problem. So in order to start a project, we're going to look up here and you can see here it says projects. So if we click on that and then we'll get this pop up to create a project, and I'm going to paste in the name of a project that relates to what we're dealing with at the moment. So this is the track suit brand spring launch, and in fact, I'm going to give it a slightly different name. So we're going to call this the Passion track suit, spring Launch and create the project, and there it is. So to make a project useful, the first thing we need to do is put some kind of a chat inside of the project that lives over here. And what we do that is literally just start chatting. So I'm going to paste in the chat I had earlier, which is we are planning a new premium track suit project will include research, visuals, product copy, and decisions. Start by asking me what information you need, and let's be a bit more specific. So I've made it. Start by asking me what information you need to reach our goal. Give me a simple bullet pointed list, nothing more. So this will help us to plan out everything we're going to need in this project. So ChatGPT has started by saying target customer profile, the price point and margin targets, the fit, the fabric, construction details, colors, size range, brand positioning, basically all the different things we need, key competitors launch timing and any constraints. So I can now now that I have this chat, this, to me, looks like the beginning of planning. So I'm going to actually rename this chat, which I do by going here and rename. So now inside of our project folder, we've actually got inside of the project inside of our project folder. Now we've got a planning conversation going on, and we can put our planning stuff in there. But also the way to move it in is to hit the three dots and then go to move project. And there you go. So now we've got our conversation inside there. And as we know it's regarding the track suit, I'm going to rename this again and call it product description. Now, if we click inside of product description, we can see that actually we've discussed a lot of stuff more than just the product description. We've talked about things to do with the images. We've generated some images in there as well, and we've dealt with product description. We've done some research. I think research is a good all encompassing name for that, so let's rename it research. And so now we've got product research and we've got planning inside there. And so now what we can start to do is build up a set of conversations all based around the passion track suit, spring launch. And then arrange it into different folders. So the project allows us to do that. And if we've got different launches, we can just create a new project for each of the launches. So let's talk about how ChatGPT uses projects. Within a project, ChatGPT assumes shared context. So you don't need to re explain the goal every time it knows what the goal is. And earlier work informs your later responses. So if you've done some work, you've done some research, you've done some planning within the same context of that project, then it understands what your goals are and what you're trying to do, which helps when you ask any questions or give any instructions. Now, conversely, and this is all within a project. Now, conversely, outside of a project, GPT treats each chat as a standalone, and the context resets unless the memory applies. So that means that in each chat, you're having a conversation. It doesn't necessarily know your goals if you create a new conversation, a new chat because they're not within the same project. And the context resets every time unless memory applies because when you apply memory and you set it up so that memory applies persistently across chats, then only in that case, does the memory remember what you've done in other chats, even then it's only to some extent. So projects reduce the setup friction and they don't reduce the quality of thinking. And what you can do is you can set up a project so that whenever you are creating a new chat, it already knows the goals, and therefore, you don't need to set up the chat from beginning and remind it from beginning. So when you do create a project, Things to avoid using them for one off questions, things that aren't necessarily going to stand the test of time aren't necessarily part of your long term goal. Avoid private or sensitive topics because it's going to remember that context and start using that for your future conversations and avoid things that are completely unrelated to that project. Sometimes we could get lazy and just use the project because it's there and we can't be bothered to click out of it and start a new conversation. However, what that's going to do is it's going to derail your project, and it's going to start remembering context about things that aren't even related to your goal for that project. So don't ask random questions inside of it because it defeats the purpose. So now it's your turn. What I want you to do is create one project for something real in your life and your career, your business. It could be a product launch, a new course module for a course you're creating, a client engagement for a client you're dealing with, some kind of admin work that needs to be done in your company, anything at all. Start one chat inside of that project that's relevant to the project, and immediately you're going to feel the difference because you're going to feel more organized. And also, as we carry on through this training, it will all make a lot more sense why you need projects and how it helps you to organize your work and organize your world. So how fun with that and see you in the next lesson. 19. How To Run Repeating Tasks Using Scheduled Tasks: So in this lesson, you're going to learn how to run repeating task using scheduled tasks within ChatGPT. Up until now, everything you've done with ChatGPT has been on demand. You ask, and ChatGPT responds, and then the work stops. Schedule tasks change all of that. They allow ChatGPT to run work in the background, repeat it on schedule and deliver outputs without you starting a chat. So in this lesson, I'm going to show you what schedule tasks actually are and how you can run them within ChatGPT. So the feature we're using is called schedule tasks. Why we use it is to run work automatically on time. And what it solves is when you want to do things like run repeating reports, which are the same thing over and over again, get reminders or get summaries of things that are already out there. But basically, any task you would want to repeat, you can do with scheduled tasks. Now the thing to remember is a couple of things. First of all, the version of ChatGPT, you're on matters, and the reason is because the way that you access scheduling changes based on going to show you on ChatGPT 5 0.2. Anything prior to 5.2, you're going to access it using the plus menu next to the chat. But I'll go through that. We're going to access it differently here. The other thing to remember is you currently need a paid plan in order to access scheduling, so you can either just follow along with me and just see what I'm doing or you can actually set it up yourself. You'll need a paid plan in order to do that. So in our demo today, creating a scheduled task, we're going to use a prompt, and what we want to do is we want ChatGPT to summarize marketing trends regularly without remembering to ask every week or every day or how often we need it. And that's going to keep our business up to date on the latest trends, and we don't even need to think about it ChatGPT will do it for us. So here we are in ChatGPT. Now, depending on the version of ChatGPT you're on, you may find your schedule under here, and you'll see the word schedule task, hit schedule or schedule tasks here. However, I'm on ChatGPT 5 0.2. And the great thing about 5.2 is, in order to schedule a task, you simply need to ask ChatGPT to do it. So what we're going to do is we're going to if I paste in our prompt. So what we're doing here is we're saying schedule a task to summarize the latest premium mens wear tracksuit trends every Monday at 9:00 A.M. UK time. So simply by asking ChatGPT to do this, it will be added to the schedule, and it will run as a task. So let's do that now. Excellent. So we can see thought for about 5 seconds on this one, and it says, Got it. I'll summarize the latest premium menswear track suit trends every Monday at 9:00 A.M. UK time, which is great. So the way to check that is if we go down here and then we go to settings, you'll see it pops up this window, and on the left hand side, we can see schedules is one of the items on the menu. And you'll see here it says ChatGPT can be scheduled to run again after it completes the task. Choose the clock schedule from the three dots menu in a conversation set up future runs. Well, actually, the way it works here is first of all, we can go into manage, and you can see, I've got one that was already here. And this new one is scheduled and it's called summarized premium Mens wear tracksuit trends, and that came from our prompt. Then it says, what's been run recently or what's been configured recently. And you can see there are two. I created this get AI New summary one as well earlier, but here's the one that we created. If we click on this, it essentially just gives the name and the instructions, which is exactly the prompt that I typed in. You can see it's set to run weekly every Monday at 9:00 A.M. Simple as that. And you can edit it also by clicking on the pencil. It pops up the same thing. If you click the three dots, you can pause it or delete it altogether. And click in here, if you click into the pencil, it literally pops up the same dialogue where you can set it. So that's the way to amend the schedule task. And this is a nice easy way to see that your schedule task has actually been set up. So I'm going to amend this so it runs now. The time now is 16:49. I might be able to get it to run for 16 50. If not, I'll change it to 16 51 and see what happens. So to do that, we'll go in, and it looks like it hasn't got 16 51. So I'm going to set it up to run 5:00 P.M. And today is Thursday. So I'm going to get it to run at 5:00 P.M. On a Thursday. Save that, and we'll come back and see what happened with it. You can also access the scheduled task here. So as soon as you run it, it gives you a summary saying that it's going to run it, and you can access the settings here as well, and that pops up the dialogue, and you can access the same drop down where you can pause it or see all of your schedules which pop up here. So just to recap what scheduled task can do well and they're not as good for, but what they work best for is weekly summaries, daily check ins, regular reminders, repeating research updates or literally anything that you want to run regularly. But what they're not ideal for are complex, conditional logic where it needs to calculate something multi step workflows, which need to contact many different components, many different nodes. That's the kind of thing you would do. In more of a workflow tool or real time decision making. That's better to be run by yourself at that time, or done by a human based on the output of ChatGPT. So those will come later with agents and automation. But in the meantime, this is an excellent way to run anything that you need automated output for from ChatGPT like these. Now it's your turn. What I want you to do is create one scheduled task using a prompt, get it to run daily, weekly, or monthly or whatever works for you, and then check the settings and schedules to confirm it exists. Once this is set up, make sure it works, and then you can be sure that a chat GPT, and it starts working for you. You can get it to run in advance, repeat tasks, and make your life a hell of a lot easier, all based on what's going on in the real world. 20. What is a Custom GPT: Lesson, we're going to show you how to build a 247 autonomous customer experience agent with a custom GPT. Now we're going to apply all you've learned in a single practical project. Let's be really precise about the language we're using. This lesson isn't building a fully autonomous AI agent. Will Building is a custom GPT that can handle customer questions, 247 consistently using your brand knowledge with human oversight, and that distinction really matters. What is it actually that we're building? So the project creates a custom GPT, and the custom GPT has brand specific instruction. So you can think of it as our specialized version of ChatGPT with specific instructions, and it has uploaded knowledge such as FAQs, plosies and tone, and customers or staff can interact with this GPT at any time. What it doesn't create is a self triggering automation, like AI automation. It's not a system that takes actions without any prompts, and it's not running in the background continuously. So when we say 247, we mean it's always available when a human opens it and asks it a question. So I want to be clear here because people use the term agent a lot. It's not an agent in the sense of a true AI agent in this stage, at this stage. It has the capability to answer questions like an agent and respond to questions like an agent. It uses knowledge like an agent, but it doesn't take actions on its own completely autonomously, and it doesn't run without prompt. And this is a foundational building block that we will later use to build a true agent. So you've got to understand that before we move also, as part of this is the concept of having a human in the loop. So this GPT is going to benefit from human oversight, and that's a feature, not a flaw. So, for example, we can use a GPT to draft a response, and then the human approves that response. We can use a GPT to flag uncertain questions, a GPT to escalate edge cases, and we explicitly design the GPT to answer confidently only when it's certain and defer to a human otherwise. And that's what makes it safe for a customer experience. So in terms of the features we're going to use in this lesson, we're going to use custom GPTs which we've already spoken about, and we use that when we want a persistent role, tone, and behavior every single time, rather than just relying on chat GPT. We're going to use knowledge uploads, which are files, and we use that when we want consistent answers based on those files, based on that knowledge. We're going to use instructions or prompts, and those are going to give us guardrails and escalation rules. So we'll tell it when we want it to go to a human to do something or what to do and what not to do. And we're going to use the chat interface as usual for human and loop control, so we'll get involved when we need to. 21. PROJECT: How to Build a 24/7 Customer Experience Agent: We are on ChatGPT, I'm not in the same account that I used for all of the other prompts that I've shown you in the lesson because we need the paid version. So I've come over to my paid version of ChatGPT. So in order to create this GPT, which is going to answer our customer's questions 247, and it's going to be a specialized version of ChatGPT. We want to, first of all, go over to here where it says GPTs and then we want to go up and click the Create button. Here you can see it saying, Hi, I'll help you build a new GPT. You can say something like, make a creative who helps generate visuals for new products or make a software engineer who helps format my code. What would you like to make? So, down here, we're going to type in our prompt. And we've said Crick customer support assistant for a premium tracksuit brand, simple as that. And then we're going to submit that. So that's going to pause for a little while while it thinks about what it needs to do. It now says updating GPT, and we'll wait for that to finish. So now you can see it said, Great choice. That's a strong clear use case. I've set up the assistant behavior to reflect a polished premium custom support experience for a high en track seat. For the name, how about Atia support? Would you like to go with that, or do you have a different name in mind? So I don't think we're going to go with Atta support. So what we're going to do now that it knows what you want to do, we'll go over to Configure, and we'll fill in this information to give it complete direction in what we want it to do. So first of all, I'm going to type in some pre prepared information. Passion sports customer support assistant, we're going to put premium tracksuit brand, customer support assistant helps answer customer questions of products sizing, shipping returns and brand policies for a premium track suit brand. In this instructions area, we can actually open it up a lot wider. And so instead of these instructions, which look pretty good, I'm going to put those down the bottom for now. At the moment it says, you're a customer support assistant for premium tracksuit brand. You represent the brand with polished friendly and confident tone, and things of that nature. So it looks pretty good, but I've got one I prepared earlier, so I'm just going to paste that in. So this says you are a customer experience assistant for premium Track suit brand. Your role is to answer customer questions clearly pull in professionally. Use only the information provided in uploaded knowledge, maintain a confident but helpful tone, avoid guessing when information is missing. If a question can be answered confidently, say you're not certain. So just contacting human support. Do not invent policies, dates, or guarantees. Never make legal promises, invent pricing, delivery times, or stock availability. Answer outside the brands. Never answer outside the brand's documented policies. So we're going to use that for now, click Close down at the bottom. So now we've got some great instructions there for our agent, for our assistant. Now, in terms of conversation starters, I think these are pretty good. What size should I get in this track suit? How do I return or exchange an order? What materials are used in your track suits? Can you help me track my order? I'm just going to change this slightly to what size can I get in this track suit? Great. Now, the next thing you can do is upload knowledge, and this is essentially what ChatGPT or the custom GPT we're creating will use to answer any of the customer's questions. So here it says, conversations with your GPT can potentially reveal part or all of the files uploaded. I've already created a document that we're going to use. Let's have a quick look at that. So here we have a PDF. It's called the passion Sports customer support knowledge base. It's got an overview of the product, product launch details, where to buy available sizes, available colors and styles, return and exchanges policy, regional return policy, shipping information, FAQs, and customer support contact. I created this before, and this should be perfect. The only thing I'm going to do is update the launch date. So let's upload this into our custom GPT, and then it will have all the information it needs to answer questions. So here we are back at GPT. We go to upload files, download, upload our PDF. And there we go. So we can always upload more information or more files here if we need to. Down here, we've got recommender model, and this allows us to recommend an AI model. So one of open AI's models to the user, if necessary. I don't think we'll need that. I mean, usually, you could say use GPT four, zero for this or 5.2 for that. We don't have any need to recommend that at the moment, dealing given that we're dealing with a sports brand that makes track suits. Here we have capabilities, web search canvas, image generation. And generally this is about what we should allow this GPT to use in order to answer questions. At the moment, we're not going to allow image generation or Canvas. We will allow web search. Sometimes customers have questions about delivery times versus Amazon or versus some other company, and we want to be able to search the web to get some context to give quick answers. So that's all we'll give at the moment. And we can review that if we need pretty much it. We don't have any new actions we want to create. There aren't any additional settings we want to set. It's currently using conversation data in the GPT to improve their models. We're helping GPT with that. I won't change it in this instance, there's nothing too sensitive in here. And that's pretty much it. So the next thing we want to do is come over here and we want to preview what it's like to actually deal with this GPT. You can see here we're using the current model 5.2. It's called the Passion Sports Track Suit Customer Support Assistant, and it's a premium truck suit brand customer support assistant. It helps answer customer questions about products, sizing, shipping returns and brand policies for a premium truck suit. Underneath, we've got some questions, typical questions. What sizes can I get for this track suit? How do I return or exchange an order? What materials are used in track suits? And can you help me track my order? So if we ask a question, we can do that simply by typing in a question down here, just like a normal chat bot. What we can do to test it is we can choose one of the questions up here. So there we go, so it's now put in our question. What sizes can I get for this track suit? And we can see that it says the passion sports Valere track suit is available in the following men sizes, XSLEcel and Dub Excel. All sizes are based on standard UK men sizing, and a detailed size guide will be available on each retailer's product page to help you choose the best fit. If you'd like help deciding between two sizes or I have questions about the fit, feel free to ask. And it's getting all that information from our PDF. So let's just check our PDF. We want to see all of these sizes in there, and it says standard UK men sizing. Let's see where it got from. Here we can see all the available sizes, XS, all the way to Excel. And up here we can see it's a Valor track suit, which is exactly what it's telling us. And it says that it's a premium men's track suit, as well. So it's definitely correct in the feedback that it gave us here, and we can see all of that information here as well. There's a lot more information here. So if you have a quick look down here in this return policy, we can see it's three returns, 30 days in the UK, 30 days in America. And 14 to 30 days in Asia Pacific. So I'm going to go back and test some of this. So let's ask what the return policy is in Asia. In fact, let's just say Asian and see what it does. And there you go. So it says, In Asia Pacific regions, the return policy for the Passion Vale Tracksu is as follows. Returns are accepted 14 to 30 days of purchase, exactly as it says in the document. So this is working really well as a customer support assistant. So once you're happy with it, click Create, and then your GPT will be ready for the world. You can work out whether you want it to be only for you, anyone with the link or go into the GPT store. Only for you speaks for itself. Anyone with the link also speaks for itself, and I would do that if you want to send it to specific people in your organization or to the public. GPT store allows you to share it literally with the world. As this is a test project, I won't do it yet, but I will create it as anyone with the link so that I can show you how you can share it with people if you want to. So I'm choosing anyone with the link. I could actually choose only me or GPT store, but I'm just going to go for this one and click SAVE. So it takes about 10 seconds to save it, and then I can copy the link, so you can see the link is copied to the clipboard, and then I can view the GPT. So when I click View GPT, you can see it's actually here under GPT. It's added a new one called Passion Sports tracksuit. That's actually the GPT we've created. So if we go down here, we can now ask it questions just like we did when we were testing it, so I can ask it what sizes I can get? How do I return or exchange? What materials are you? So now that we know our GPT works and we can see it in the menu, we can go off and access it in anyone in any account. So you need to be logged in to do it. Once you're logged in, if I go over to another account, I can show you exactly how to use this. So here I am in my other account, my product account. And what we're going to do is we're going to access the GPT as anyone could. So the way we access GPT is essentially just paste in that link, hit Enter. And you can see it's opened the passion sports tracksuit customer support assistant right here. It says who it's by, and it's got everything we need now to start asking questions. So now it's your turn. What I want you to do. First of all, step one is define the audience, decide who this GPT is for. Is it for an internal team, for customers, for students, for yourself? Work that out. Then step two, I want you to finalize the role. So what is this GPT doing? Mine is a customer support assistant. You need to ask yourself, what should this GPT answer confidently? When should it escalate to a human? What should it refuse to answer? And then step three is actually create the GPT using the instructions I've given you. Use the configure screen, add clear instructions, upload only the stable knowledge that it needs, and enable only the capabilities you actually need. Step four is share carefully. Start only with yourself, testing it. You don't need to send it to anyone. Once you're happy, get a link and control the sharing by sharing that link only to the people that you think need to test it or need to be aware of it, especially if there's sensitive information in there and only set it to go public and maybe put it on the GPT store when you're confident in its behavior and you want to share it with people. So the key takeaway for this project you've created is that this custom GPT isn't just an autonomous customer support that acts on its own. It's a 247 first response customer experience layer. So it's customer support that can reply to people 247, always with a human in the loop when necessary. And that's exactly what makes it useful, safe, and teachable, because that way you can get the best of both worlds. You can answer questions for customers about any product or service or answer anything internal to your company, whatever you think suitable and keep a human in the loop at all times. So there you go. Have fun with that, and I look forward to seeing your GPT working for you. A