ChatGPT for Marketing | Learn Generative AI: Prompts , Strategy, AI Content Creation, Copywriting

1. Welcome to Generative AI for Marketing Training: Hello, and welcome to this practical guide on using Cage Pty for marketing. My name is Leo. I'm your instructor for this course. So just a few words about myself. I'm a full time AI product manager, currently at a global enterprise with over 10,000 employees. I've been in sales and marketing since 2012. Generative AI and CGI PT in particular, have been the biggest game changers in this field. It's in every conference, every report, every meeting. AI is already changing the way we think and work. And it's also affecting the way we hire and are hired. So if you work in any marketing direction, whether it's social media marketing, content marketing, digital marketing, SEO, copywriting, or small business entrepreneurship, you will definitely find something for yourself in this course. We'll start with the fundamentals, learning how to improve GPT responses using prompt engineering techniques. Then we'll turn GPT into your marketing strategy assistant. For example, we'll be creating a user personan from a real survey and analyze competitor reviews and finding some patterns there. But instead of saying a lot about what we're going to do, here's a quick look at one of the tools that we'll use to automate a huge part of your research of information. We'll be using tools like perplexity AI. This tool will understand your request, ask you follow up questions, and turn all of this into series of searches. By the way, it searches not only the web, but Videos, as well. Then perplexity will analyze the sources and give you an answer with references and further follow up questions. And that was just a part of a two minute lecture in this course. But the most exciting things are still ahead in the further sections. We'll learn how to generate creative ideas and create AI generated text that doesn't feel artificial and is close to your natural tone of voice. We'll even have a look at the components of artistic writing. Lastly, I want you to make sure that you get the most out of this course. I've created an easy to navigate and easy to copy and paste prompt library with all the prompts and approaches that we mentioned in this course. You can simply add it in your bookmarks and bin it next to your chrome tab. Just like this. Also, if anything gets you confused or you need any help, feel free to share your questions in the Q&A section or send me a direct message. I'm also happy to connect on LinkedIn. I'll be available to personally respond to everyone in this course. Start the course today and get a lifetime access to it and all of its future developments. I'm sure that they will help you stay ahead in your career. 2. Choosing an Generative Ai Large Language Model in 2024 : The competition in the AI world is super aggressive. C GPT by Open AI made a huge breakthrough making generative AI available for the masses. But now, there are some decent alternatives, and the choice isn't that obvious. You will see a few lectures that use GPT 3.5 as this was the best option at the moment of recording. But after the release of GPT four, I recorded all of the lectures using the most recent model. I know that many of you still use the free version of GPT, the model 3.5. And now, since the competition grew so much, I would actually recommend something else, something much better than the free version of CGP. So let's first talk about the free versions. The free version of Cage Pat has only one advantage for me. And this is its ability to apply custom instructions. Basically, this enables you to enter a prompt that will describe the context about you and about how you want Cap to respond. My recommendation so far is to stop using the free chap. To either use the GPT plus account with all of its most recent benefits and features or try one of the following alternatives. Let's start with the first one. Google Gemini. Google Gemini is often overlooked by many professionals, and in my opinion, it actually has a few amazing advantages up to its leave. It's number one advantage is its integration with the Google ecosystem. Gmail, YouTube, Google Pixel phones have tons of features and benefits that you can use even in this free version. This ability has been rolled out and shut down a couple of times here and there, but in the end, I believe it's going to be the long term advantage of Google. The other advantage of Google Gemini is that you can check its responses with Google Search. Gemini will break down new response into pieces and separate searches and find search results that can confirm the chat response. I also like the user interface more than CDPT In the free version, you can also upload your image and ask to create a text based on this image, which is kind of cool as well. The quality of responses in the free version, in my opinion, is also a bit better than in GPT 3.5. Let's talk about the next one Anthropic clot. I love anthropic cloud for many, many reasons. The free version sounds much more natural than GPT 3.5. It offers the largest token sits on the market at the moment, which means the best memory of all the three models. You can use it to run longer prompts, process larger documents, have a longer conversation throughout which Cloud will remember the context. So if you summarize a two hour long video from YouTube using a third party tool like HarpA AI, for example, Cloud is often the best model to choose for this kind of task. The other interesting advantage of Cloud is that it has an incredible OCR, which, by far is the best out there. But dictionary here, OCR stands for optical character recognition. If you want to ask questions about an image or create captions for an image or decipher a written handwritten text, having a great OCR is a huge advantage. If I had to choose one free large language model, I'd definitely go for cloud at this moment. In fact, I still use it from time to time despite using a GPT plus account. By the way, the recent family of models called Cloud three claims to outperform GPT four in some use cases. And amazingly, the second most powerful model they have, which is currently Cloud three sat. It's available for free. Yes, you heard it right. It's free. But talking about the paid versions, GPT four still offers great things. GPT four offers custom GPTs and a few more nice features that make me stay with the GPT so far. Right, let's wrap up. If you use a free, large language model, triclot or Google Gemini. And ditch that free CPT. If you're looking for a paid motto, consider GPT four or cloud if talking size is the most important thing for you. All of these things are fine, and approaches from this course will work for you unless they don't require a specific functionality like custom PTs, for example. That's it. See in a few seconds and cheers. 3. How ChatGPT Works: Capabilities and Limitations: In this lecture, we are going to discuss Chagipt, what are its strong and weak sites? Chagpt is a natural language processing chatbot driven by generative AI technology. It allows you to have human like conversations and actually much more Gp can answer questions and assist you with tasks such as composing e mails, essays, and even writing code. If you dig really deep, it can be integrated with a lot of other tools to assist you in more tasks. Here are some of the downsides. The first limitation is that Chagipeti is not a reliable source of information. As of now, Chagpti doesn't provide accurate sources of information that it uses. It was trained on billions of information entries, but it doesn't know what these entries really mean, Which of these are more accurate and which are less accurate. Moreover, you don't have much control over the generated data. Each regeneration would result in a slightly or even significantly different answer, which is actually a blessing and a curse. That also means that the response that you'll get might be different from the one that you see in the demos of this course. Even if you enter exactly the same prompt, you might think, Leo, they already have this browsing feature. You can browse with Agape well in 2024, this feature is still unstable. Although Cha GPT now has this browsing functionality, it's not currently very good and accurate at searching and processing information. It often produces mistakes. Blends them with hallucinations very confidently that it's hard to notice it. I expect some improvements in the future, but that's what we have right now. For the time being, I would recommend using research specific tools like being Perplexity or U.com for any tasks that involve browsing. The second downside is a continuation of the first one. The confident mistakes that ChagiPT makes are called hallucinaus. You see, Chagpt is trained to write using a language, many languages actually. It can produce faulty information very confidently without you even noticing it. It can generate non existing sources of information that look very, very believable. It can even pretend it knows the content of a link, but in fact, Ch GPT is just guessing the topics within the URL slug. Here's a high level, but very practical tip, your good use case, no hallucinations, bad use case, lots of hallucinations. An offensive prompt wouldn't fix a bad use case. The next downside is its lack of emotional intelligence. Gbt can simulate natural conversation, but it lacks the emotional and real world experience. It's the intelligence of a human conversation partner. Gbd can have the difficulty understanding and responding appropriately to subtle nuances in communications. But hey, we humans, we are sometimes even worse than this. Let's move on. The next downside is that Jip itself is not very good at maths. At first, this was a huge problem, but now GPT users can access some of the improvements, for example, by triggering Chagpt's advanced data analysis functionality. Or by using, well, from a custom GPT. The next drawback is about privacy and security concerns. The use of GPT requires the exchange of data and information with the system. This poses potential risks in terms of data protection and security. It's important to take appropriate security measures to ensure that the sensitive data is protected and doesn't fall into the wrong hands, which actually already happened. Number six is the computational cost. Gp is a highly complex and sophisticated AI model. It requires substantial computational resources to run. Organizations should carefully consider their computational resources and capabilities before using PT. Let's say on premise, I believe this problem is temporary and with time we might be even able to run large language models offline on our smartphones. It's actually something that Apple is working on right now. But at the moment of recording this lecture, it's far from being in production. By the way, Sam Altman, the founder and CEO of Open AI is actually looking for energy resources to scale the product to new levels. So like most other AI technologies, TGP is great at finding patterns and analyzing data that you provide. Ch GPT does way more than just writing social media and blog posts. It can provide almost an infinite amount of ideas and points of view in a matter of seconds. You just need to prompt Ch GPT write and be aware of its limitations. The advantages actually outweigh the disadvantages by far far. One other advantages that Chachipt users don't utilize enough is it knows professional frameworks, whatever copyrighting formula you have, whatever format of business or marketing analysis you want, it can handle it. The primary advantage of ChagPt over other large language models is its simplicity in customization and personalization, and it's very easy to integrate in your workload. You can set custom instructions, custom puts, and incorporate Chgpti with other tools, for example, Harpo U.com Descript. All of these tools utilize Chagipt to enhance their multimodality. By the way, multimodality is an important term in Eni. It essentially refers to processing various types of media such as text, audio, video two D images, three D images, and so on. The list can go on, but it's important to remember that Gp is just one tool in your huge professional toolkit. The most important tool is your critical thinking and your experience. Yeah, let's move on to the next lesson where we will look at this tool in more detail, see in a few seconds. 5. Data and Privacy Settings in ChatGPT: Hi everyone. In this lecture, I want to draw your attention to an aspect that I'm sure many often overlook. Something that even a few employees at Samsung have done. By the way, check out this story on tech crunch. Just Google Tech Crunch, Samsung GPT band, something like that. And you'll see the first link. All right, so if you're like me working in an enterprise setting, handling corporate information, there is a crucial setting that we should remember to utilize. Particularly when you work with data that is confidential or at least close to confidential. Here's how to ensure at least partial confidentiality within chap ten. First, you'll need a GPT account while working with confidential data. Then within chat GPT window open settings. Then in data control step, disable chat history and training. The reason I say partial confidentiality is that the data is stored on open AI servers for 30 days. Anyway, I'd suggest removing any brand mentions, sensitive data, or any data that can identify your business and you'll probably get away with it if you're using a free version of Cha GPT, I would recommend to not enter any confidential data at all. It's also related to Bin Chat and Google, Gemini, and Claude. As these large language models use your input to improve their models. I know you came here for other things, but safety comes first. So please be safe and never underestimate the privacy as it starts with you. All right, cheers, and see in the next lecture for stuff that's a bit more fun than data and privacy. 6. Using ChatGPT Global Memory: Come back. Today, let's talk about an important update of the memory feature of HGPT. Open AI introduced a global memory update, and unlike the selective memory we discussed in the previous video, this type of memory can access your past chats. So now you basically have two types of memory, and you can turn off or on each one separately. And depending on what you're doing, you might want to turn these types of memory in the personalization settings of HA GPT on or off. The selective memory, you can remove each particular memory, and in the global memory, you can either enable or disable this feature completely. So it either has access to chats or not. So what is the global memory? Global memory has access to all your past conversations, but it prioritizes the more recent conversations that you have. Let's talk about what we can do with this feature enabled and why should we enable it? Think of using global memory as self reflection. It's cool to play around if you want to create a personal brand based on the highlights that GPT can find about yourself, based on your past conversations. But you'll need to provide additional important information about yourself because you probably communicate with HAGPT differently than you do with people in real life. Who doesn't know a lot about yourself. Particularly, pay attention to the emotion that you'd like to sincerely convey in parts of your task. As hagipti most likely wouldn't be able to pick this up. You probably don't joke with age pity. You probably don't express a lot of emotions with it. Okay, let's talk about some of the more practical applications of using this self reflection approach with the global GPT memory. So the first one is to create your own tone of voice DPT. You can create your own tone of voice DPT that will rewrite your text in your style, the way that you actually communicate. So the way to do that would be to ask ChaGPT to analyze the style of communication based on your past behavior, choose the types of sentences, the length of sentences, the emotional tone that you convey so on. Add a few details from yourself and then pack it into Custom GPT. Of course, in this course, you'll learn how to create a custom GPT. Don't worry about that. It's really easy. But apart from just going to text tone of voice, you can create visual assets that complement your personality. For example, you can create backgrounds, decide on color choices, font choices, styles, icons, elements, overlays, generate those in image generation of ha JPT and collect this into personal brand boook and personal templates. Also generate ideas for key messages that you will spread on social media, for example. Of course, it's more of a personal use case, but you can get personal recommendations like movies, music or recommendations related to your hobbies. Like almost every tool, it does come with certain limitations. So let's talk about the limitations of global NGPT memory. So while making my research in this topic, I found many obviously GPT generated videos that seriously overestimate the power of this feature. And suggest use cases like customer support that will pick all the history of your customers, coaching that is directed at certain learning. And realistically, currently, this feature is not even close to the consistency that's required for these types of use cases. And really a significant change in technical approach is needed to actually make it work with interface like HGPT using this memory feature. I can see HAGPT moving in that direction, but from slightly different technical angle, they're working on integrating data sources for enterprise for sources like HubSpot CRM, SharePoint, Google Drive, and so on. Let's get back to the limitations of the memory feature. It seems like with a greater increase in productivity that AI gives us, we keep switching contacts more and more frequently. That's where global memory could become a false friend. When you use it in this environment, you'll see the details of unrelated contexts blending when you don't want them to. Therefore, I would suggest making use of custom GPTs of projects, custom instructions, more and more when you have these project focused tasks where you need to get those particular bits of information more consistently. You don't need this self reflection, just switch this global memory feature off. So the next limitation is gradually evolving from the previous one, and this is the detail accuracy. It's difficult to get the memory to pick the right small details correctly. CHGPT memory prioritizes memories from the more recent chats, which means that the context is always moving. And while it's convenient for many, many everyday use cases, it's also going to introduce a bit of unpredictability to the behavior of the model. One more limitation is that this memory isn't infinite. That's probably the reason why the previous one works like that. Open AI doesn't disclose the exact memory size, but it's definitely not infinite. At some point, you'll probably see something like memory full. If you don't see that message, your older memories will fade. Not really cool. Let's discuss some of the tips. Turn the memory on when you need it and turn it off when you don't projects and custom GPTs for particular projects. The next one is important. When HGPT gives you an answer that you like that you'll use as a base for your following task, provide a feedback, even if it's a positive feedback. Typically, we just provide negative feedback. But just say something. Thanks, I'll accept approve this answer, or optionally, you can even add your final text after your edits, saying, Hey, I accept this answer. This is my final text after reviewing it. This way, HGP will learn more about your priorities and choices, and memory will have better data for self reflection prompt. Set for the second part of our ha GPT memory feature and see a few seconds in the next video. 7. How to Use ChatGPT Canvas Feature: Hello, in today's video, you learn about the HGPT Canvas feature. This feature has become standard among major language model providers like HGPT, Google, Noon, and others. It allows you to work directly within the AIS response, and it's perfect for long form writing or code editing. So what is Canvas? Let's say you're writing a feature description article. Once HAGPT generates the content, you might want to edit just a tiny part of it. And instead of regenerating the entire response, you can open it in Canvas straight in HGPT without leaving the interface. Canvas, once it's open, you can manually edit the text or select specific parts of the output and prompt only those parts just like this. Canvas is especially useful for long form content creation and also for tasks that involve editing and formatting and, of course, coding tasks. However, if you work with something really long, like a couple of thousand word length, break down the canvas into parts to avoid getting Canvas crushed. There were actually a few cases when Canvas wasn't able to handle this type of long context and failed. That's a pitty. So how to activate and open Canvas. There were three ways to activate Canvas feature. The first and the most intuitive one is to click the Pencil plus star icon. If you see this icon next to the response, just click it and Canvas will open. The second way is to type open in Canvas. If you already have the response and the icon doesn't appear, just type and chat open in Canvas, and hat GPT will open the response in the editable view. One more way to activate Canvas is to use slash commands. Press slash in the prompt field while typing the prompt and then select Canvas from the drop down. Pretty much it. Let's recap. Canvas. Let's you edit hat JPTs responses either manually or by prompting specific parts. You can access it by either clicking the pencil icon, typing open end Canvas once the response is already there, or use slash Canvas Command while still prompting. That way, the response is going to be open end Canvas. And that's it. Stick around to see a quick example at the end of this video on how the Canvas is being used in real time or skip straight to the next one. 8. How to Set Up ChatGPT Projects Feature: Hey, today's video explains the chat GPT projects feature, how it groups chats and files so we can work in a tidy space while still avoiding repetitive parts of your prompts. So the thing is the chat list gets messy very fast, and the projects feature, let's see sort the related conversations and reference files and written instructions all in one spot. So of course, even if a chat isn't in a project, you can still search chats using this search bar here, and typing the keywords, you remember in the search bar, and in many cases, that might be sufficient. However, it doesn't scale, and it's not that easy to see multiple chats on one screen, and you don't have the capability to keep the same instructions and files this way. So I would say having chats dedicated to a single project helps much more, especially if you work on multiple projects like working on multiple Instagram accounts or multiple clients or separate work stuff from personal stuff or separate different hobbies that you have. So how to create a project in HGPT open the web or mobile app of HGPT. Unfortunately, project creation isn't currently available in the Desktop ABA, either on MacOS or Windows. Click this plus icon and then name your project. Then attach knowledge base files and add instructions. Actually, those instructions are very similar to instructions for custom GPTs. Scores, there is more information on creating instructions for custom GPTs and even a GPT for creating custom GPTs. But here are a few key tips to get started right away for knowledge based files, aim for the simplest formatting possible, preferably Doc X format, TXT, or CSV format. If you use PDF, make sure it is very simply formatted, doesn't have multiple coms. It's just all single column text. That's if other options that are simpler aren't available. What helps the system read the information more accurately is making sure that information in your files is clearly formatted with all the understandable headlines and namings. And of course, you can use HGBT to help you format your input files for machine scanning without all this fancy prompting. Regarding instructions, just provide a detailed overview of the project, its purpose, who's working on it. What's the process? What are the tasks that you're solving? And what are the other rules that apply to all chats within this project? By the way, beware of this one important limitation. You cannot use custom GPTs directly within a project. I assume that technically that's because it's difficult to Ort instructions within custom GPT versus instructions in projects. One more limitation. In the desktop app, for some magical reason, you still can't switch models within projects. So I would recommend to start working with projects in the web app, especially when the model choice is important because by default, currently, the model running is GPT for. All right, let's summarize. You can store your chats within the single folder, and you can also add the instructions and files related to single projects so that you don't have to repetitively enter the context. And if you want to use different models within projects, use the web pp. If you want to use Custom GPTs, start the chat as normal, use the custom GPT, and then add the chat into the project. That's it for this video and see in a few seconds. 9. Meta prompting: Enhance Your Prompts with AI: Hello again. Today we cover metapmpting. It's a crucial skill now and it's super effective for interaction with AI. So what is metaprompte? Long story short, it's using prompt to create or refine another prompt. So you prompt an AI to help you write better prompts. More like prompt assistant, I'd say. So why use metapmpte? Well, first, it leads to sharper prompts, and this is an obvious reason to do that. You also get faster results and more consistent outputs. Also allows you to instantly use advanced techniques, and you learn the prompt structure by seeing how metaprompt refines your inputs. And speaking of your inputs, let's talk about when you should use the metaprompte. One of the first cases I would recommend and it's typically used in those cases is for complex tasks that require nuanced instruction. One more reason to use metaprompting is to refine raw ideas and turn them into polished prompts. Particularly how I like to use it is I actually use a lot of voice input and dictation because I speak a bit faster than I type, and also I also kind of think more naturally when I speak. So I've built a tool for you. It's called LP Meta answer. It contains tons of research on prompt engineering condensed into a knowledge base that is optimized for the machine scanning, and it has all the principles that we cover in this course. So it's also going to be useful for you to use the knowledge that you've gained in this course and apply it to your real tasks in real world. So updated regularly, and I plan to extend the range of tools that are supported by this prompt hanser. But right away, it already includes all modalities. It includes techniques for image generation, video generation, deep research. So all of these types of prompts are supported here. Now, here's how to access and use it. Use the provided link to access the tool. It's available in the notion resource. It's available in the course as well. And if the link is broken, please reach out to me either via LinkedIn or private messages. But even if you use hATGPTFree tier, you can use it here. In HAT GPT, you'll find the custom GPT named LP Meta Prompt enhancer. Once you've got it open, there's one critical first step. Your very first message in this custom GPT must be exactly this. Enhance my prompt. Deep research prompts, start with enhance my deep research prompt. This specific input in the beginning of conversation is necessary because it activates the correct next steps within this custom GPT. Basically, it allows the GPT to review the extensive knowledge base I've provided and also tune in to the task of enhancing the prompt instead of executing it. So after sending enhance my prompt, past your prompt draft. Either it's voice node or any raw context work, this is particularly effective for unrefined ideas or initial draft. Tool will then provide an enhanced prompt. It will also explain the techniques it incorporated for the enhancement. And I did this step particularly for your better learning experience. And yes, this enhancer is versatile. You can use it for text generation prompts. It also works with prompts aimed at music, image and video generation. However, for particular tools, if those tools aren't in this GPT, you may want to look for specific custom GPDs. For example, if you want to navigate a huge tool like Sno AI to create music, I would recommend to go to GPT store and search for custom GPDs called Sno AI. Play around with that. But for learning experience, this tool includes all the techniques discussed in this course plus more advanced material as the research is growing. But keep in mind that its best use is to enhance the input that you already have. Because if you don't have the value in the input, and you don't have the right task, the right context, then it will be just putting text from empty box to another empty box. But you can also try and generate a prompt for a specific case. For example, you can paste text over an article ask for five image generation prompts for its banner, and you can even generate a few ideas for those prompts or share yours. Of course. Also, there are two steps of the enhancement. The reason for that is to give you more flexibility and not to overdo with the enhancement or steer away from the original intent. So once the prompt is enhanced, the system will suggest enhancing it even further. If you want to experiment or have more details in the prompt, just say yes. And quick note on the troubleshooting. Occasionally, GPT might try to execute your draft prompt directly instead of enhancing it. If that occurs, regenerate the output. Alternatively, if it keeps happening, just place your entire draft of the prompt or voice node in quotation marks, just like this. If that doesn't help, type prompt first, then Column then put your prompt into the quotation marks. This signals the GPT to process the text and enhance the prompt and not execute the prompt. Once you receive the enhanced prompt, just copy it. Open a new chat in Chat GPT, choose the appropriate AI model for your specific task, and then paste the enhanced prompt. For certain use cases, make sure you attach any additional notes, documents at the end for optimal results. You can even attach your voice input as a context dump in the end of the prompt. So, yeah, give this meta prompt enhancer a try. Explore its capabilities. Experiment with your prompts. If the access link doesn't work for some reason, please drop me a message here or on LinkedIn, and I will update the link and share the updated link with you. Because sometimes these links can be updated by Open AI without any warning. I don't even know about it if no one tells me. Now, let me turn on some music and enhance a few prompts to demonstrate the workflow in real time. Feel free to boot on pause and read into more details or skip and play around with your own problems. Seeing the I'm using Google Wisk because I really like the Image four image generation model, but this will also work in most other image generation models. 10. CHATGPT FOR MARKETING STRATEGY: Marketing isn't creating an Instagram account, publishing post, and just launching ads. Marketing is a form of customer centric management that generates value for consumer and profit for business. To do this successfully in the 21st century, you need to navigate in the systematic analysis of markets and statistics. Apply empathy and creativity and do this fast like today. So depending on your organization, you'll have different budgets and time to create a strategy. But what if you need to quickly come up with a zero budget strategy by tomorrow, invalidated as ECO alarm. Of course, I'm exaggerating here a bit, but I'm sure you know what I mean. In this tutorial, we learn how to quickly find and process publicly available data, put it into actions. We'll learn frameworks and even create our own GPT with strategy that will respond to your prompt in the alignment with your strategy. We'll cover searching and collecting information and data. We'll cover competitive analysis, target audience definition, marketing, and content mix, and much more. By the end of this tutorial, you will learn not only to copy some prompts that you find online or copy my prompts, but also a set of approaches that you can extend and adapt to your context. Alright, I hope you are as excited as I am, and let's get started. 11. CHATGPT FOR MARKETING STRATEGY: ChatGPT for In-Depth Marketing Landscape Analysis [Macro]: Hi and welcome back. In this lecture, we will give tragic beauty, its first assignment in the strategy workflow. And this assignment, we will give chat Jeopardy. It's row. We will set rules on what it can do and what it cannot do. And we will also give the market data. It's important to start with this step if we want to proceed with other analytics. And the reason for this is that we want to make sure the deputy understands the context. People on Udemy learn at different paces and have different levels of awareness and the topic. I decided to try the following format. First, we will run very quickly through our process. And then I will go step-by-step in a more detailed instruction. This way, if you're familiar enough with the topic, you'll quickly get the result that you need. And if you need some more guidance, you will have those steps to. So if you already understand what's going on, feel free to fast forward the lecture to the very end and move on. So let's start with a quick run through. So our first step is to feed, check GPT worth the market data and process it. For that purpose, you might either want to prepare your market insights that they already have or go to Bing Ai chat and ask it to do the research for you to do that, simply go to the Edge browser, open a chat based the prompt that you can see on the screen right now, then copied the result that you have. Go to check GPT, enter the prompt from step number two, the one that you can see on the screen right now, enter your industry, company and paste the summary from step number three. Expand the result that you have using PESTEL analysis. One more optional step that you might find useful if you want to go to a new market and analyze its attractiveness, is to elaborate using Porter's five forces model. The prompt for it is on the screen right now for you. No worries. If you haven't done any of those analysis before, you will find the reference to these terms in the end of the section in the supplementary materials. Now let's go step-by-step. So now we are at my screen. I've opened Microsoft Edge browser, went to bing.com, logged in and entered the chat. I chose the general tone of voice. But you could also try precise if you want shorter answers. I wanted to keep some of the details but not overwhelm tragedy. I pasted my prompt. Here. I added the row. I gave the task to research data about the industry. In this year. I also suggested some of the reputable resources it can try, but I don't want to limit, I just want to start looking from these resources. I also don't want bank to make things up or give data without any references. And I also want to specify the volume of analysis. In my case, it's tan points. Let's have a look at the answer. I can see that it has given us the answers. I can see Deloitte as a report. I can see some other websites as well here, but it's quite detailed. Let's go and copy this response. Or next step is to go to Chad GPT, paste it in our prompt in the end. So here we are in charge of beauty. But this complete prompt built here. So I assign the role to tragedy as well. I assign the industry and company. I mentioned what I will provide and I give the task to judge between. I also give a restriction not to add any sources or make anything up. I also instructed to not use any other sources. Then I paste the trends. So you might ask, why would you ask to summarize what's already summarized? The thing is, I won't charge a pity to process this information as it's own input to really understand and to keep these factors in mind when working with the future desks. Let's think of this as a step zero. What I'd encourage you to do here is to verify whether all of this information is correct, read it carefully, and move on. The next step. I suggest It's taking the PESTEL analysis. This is a typical analysis used in the early stages of marketing strategy. It's important to analyze the environment. I also asked to use the table format. So let's have a look. What we have here. We have the factors, the political, economical, socio-cultural, technological, environmental, and legal. We have the description of each factor that influences the market and we have the impact on the company. I can see that in the description it uses the same arguments that were in the previous summary. And the impact on Ford Motors also sounds quite reasonable. That's it. Let's summarize. In this lecture, we have learned how to research the market with AI, how to feed your market data into judge, deputy and set the marketing landscape. We also expanded it using PESTEL analysis. And now we're ready to move on. But then analyzing target audience and competitors. See in the next lectures. 12. CHATGPT FOR MARKETING STRATEGY: Identifying Target Audience, Segments, and User Persona: Hi and welcome back. In this lecture, we're learning how to define and segment our target audience. We will learn multiple approaches. But which one is best for you is up to you to decide. Feel free to modify them. Skip some steps if you want. No one knows your case better than you do. So let's transfer to my screen and have a look at the prompts. So when you know your target audience, your prompt sounds like this. Outline the main target audience segments for company. Create a table with audience demographics, desires, problems, goals, behavior patterns. Elaborate on these segments, and then you use semicolon and your segments. It's very important to know your audience. So if you already do, this way, is faster. But let's have a look. What if you don't know the audience yet or at least not in this detail. In this case, our prompt is going to be a bit longer. We define the role as usual. In this case, it's the same, but let's keep it defined. Let's ask tragic putty to pretend that it knows everything about every market. Outline five audience segments for the following category, and then indicate our industry. The amount of segments could be different. I started with five, but it also could be for three or whatever. Let's do five because it's more or less realistic. Each segment should be defined by a deep, underlying category related need. Give each segment a title that will describe the audience according to the marketing strategist Standards. Describe each segments, category related needs, pains, desires, and demographics, age, gender, marital status. You can add any other information that's critical to know about this audience. You can't describe the same audience Twice. Use a table format. Let's have a look at the results that we have received. So why can see five segments, family-focused drivers, luxury car enthusiasts, eco conscious consumers, commuter, convenience seekers, and adventure seekers. First of all, I must say that I really like the naming. You can already imagine that person just from the second title. Let's have a look at the needs. Family-focused drivers are seemed to need safe and spacious transportation for families with children. Pain is lack of safety features in space current vehicle. That sounds quite realistic. Of course. My recommendation is here to go and refund those target audiences, the rock qualitative and quantitative researchers. And if it's a big company with a budget for research, That's already something that you can do, but that's a zero budget research. There are other ways to work around it. That's not the purpose of the course. I recommend to verify these results with real human consumers of your product. But this already gives you an idea of who to ask, what to ask them, which problems to verify, you can already build some hypothesis. Isn't it cool? So let's assume that at the moment, we already have our target audience segments verified and approved. Next is one of the approaches that was partially covered in the previous table. It's Kotler's segmentation approach. There are many things that Cutler has described about target audience segmentation. But in this case, we are talking about demographic, geographic, psychographic, and behavioral segmentation. Our prompt sounds like this. Expand using co-authors segmentation approach, demo, Joel, GO TO cycle behavior, use a table format is all segments you mentioned in the previous response. You can also ask Judge deputy to add a typically preferred product in this category. In this case, we would ask at a typical car choice, I will do it in the separate prompt. As you can see, these responses do not contradict our previous table, but we got a slightly more expanded table here. But what do we do with all of these tables? If you have spent some time trying different approaches for target audience segmentation, feel free to ask Judge ability to merge two tables into one and it prompted to do so should sound like this. Now summarize by emerging the previous two tables into one. And here is the moment when I'm adding one more column C on which car models each segment would typically prefer. You can summarize, you can't duplicate findings, use a table format. Now here's the result. As you can see, all the typical car models are not super correct. I mean, they are only Fords. And while that might make some sense, I think that would be more useful to have other Garmin factor models as well. So I'm asking to rewrite the last table. And typical car modal don't limit yourself to only use popular choices by competitors team. And now we have it. Now that looks correct to me. I can see EVs and eco conscious consumers, B and C class sedans and commuter convenience seekers. That all makes sense to me. So this is actually already usable content that you can add to your prompts when you ask to create something. And this will help you come up with a more relevant content. But we can go even further in our understanding of the target audience. And we can ask chat GPT to create a user persona if you haven't spent years completing such tasks. It will also make you a better marketer because it shows you the exact form, fields and types, and examples of this approach on your case, on the one thing that you are working on, this connects to your experience and you're not only accomplishing some results that are actually getting paid for, but you're also developing as a professional. So don't be afraid to use some techniques and approaches that you haven't used before. It's actually a great way to learn the new stuff. And if I'm not mistaken, hasn't really been so accessible in the past. So now that we've done all of those complicated prompts, you can simply asked to create a user persona for particular segment. And you can see so much more detail about this person, e.g. here's Alex, adventure seeker, age 28, and gender, male, marital status, single. He's a freelance graphic designer. His interests are hiking, mountain biking, snowboarding, rock-climbing, camping, and music festivals now has behaviors, goals, and motivations, pinpoints and frustrations, buying motivators. So you can really dive deep into what these people are. Always remember to do the reality. Check, if you have a zero budget research, find someone your network in your surrounding who has a similar behavior and talk with them, observe what they actually do. So again, it's important to remember about this reality check. Now, instead of asking for each persona separately, I'm just asking, great. Now create one user persona for each segment left, and we have all of them written out here and minutes, maybe even less. Alright, let's summarize. In this lecture, we have learned how to elaborate on the audience that you already know, how to segment it and create user persona. And also we've done the same for audiences that we don't really know very well yet. But we've laid some ground for validation in the future. We've learned how to create user persona. We've learned how to use couplers, segmentation approach, and a couple of other segmentation techniques. Target audience information influences all of the future steps that we take while creating content, ads and tactics. Hope you liked this one and see you in the lecture on competitor research and analysis. See you there. 13. CHATGPT FOR MARKETING STRATEGY: AI-Powered Competitive Research and Analysis: Welcome back. In this lecture, we are learning to work with competitors using tragic beauty. By the end of the lecture, you will know how to use judge a putty properly to collect quick information on their competitors, who they are, what are the links to their pages? And some basic open like public information. You will learn how to analyze and summarize bulks of customer reviews and how to use both marketing landscape and customer reviews to create swat, tows analysis. So let's jump right in with the quick walk-through. And then we will go through each step in more detail. Our first step is to collect competitor data. We're not talking about market shares and complex economic data that's not publicly available. Well, we are talking about simpler stuff. So let's ask Chad GPT-2, keep acting as a marketing strategist. Create a table with ten someone's competitors. In this case, it's ten. Ford's competitors include their website, Instagram page, rank, country, and most popular products. In this case, these are car models of each competitor. Step number two, analysts reviews. We need to go to Bing Ai chat, write the prompt that sounds like this one. Analyze 200-year use of Ford Motor Company. You'll analyze them and find common patterns and summarize and pros and cons. Table mentioned how many consumer reviews featured each argument. Use my source. Let me quickly explain you why. First, I want to find the source that represents the type of reviews that I am analyzing. Second, I want to know how many reviews being has analyzed. And also I want to know how many people mentioned the same thing. I mean, people mentioned multiple things in their reviews. So it's important to know how each advantage or disadvantage influence the general picture. Then we copy the results, go back to Egypt T and ask it to create swot analysis, including the information that we have analyzed previously. And then you information that I will provide. This prompt sounds like this. Create a swat analysis for Ford Motor Company. I will provide you with a consumer reviews summary. You'll create a two-by-two table for the analysis used both PESTEL analysis above competitor data and consumer review summary. Here is where I paste the summary. Alternative way to create swot analysis is to ask to create it in relation to your company. This part in relation actually set some more context to charge. It's interesting how it really compares these two things. So first, create a swat analysis for your company in relation to another company. So create a swat analysis for ford in relation to, let's say, Toyota. I'll provide you with consumer review summary. You'll create a two-by-two table for the analysis. Use both PESTEL analysis above competitor data and consumer reviews. In this case, I will need to go through the procedure of review summary twice as the next step for the swat analysis is the analysis that I call. So what? Because it really gives us a bit more insight into what to do next. I'll show you more details on this later on in this lecture. So basically the prompt sounds like this. Now expand using toasts, analysis, strengths, weaknesses, opportunities, and threats from the previous message. So it's basically everything we are doing throughout this lecture. We're collecting competitor data, analyzing reviews, grading swat analysis, and then expanding using toasts analysis. Now let's go through all of this step by step in a bit more detail. And I will show you what Teddy PT, and Bing Ai provide us with. Let's have a look at what we have when we asked to collect their competitor data. So here's our prompt. Keep acting as a marketing strategist. Create a table with 104 competitors. Include their website, Instagram page, brand country, most popular car models. I can see here that we have the country information, correct? I mean, everything looks correct. The websites look correct as well. However, I can see that some of the websites are localized for the United States market. So always we find this information. If that's not the market that you need, you might want to go and change this link. I mean, if you copy this table into your spreadsheet, then it's easier to quickly change this information. I can see that whenever there is a localized page, e.g. on Instagram or website, it chose the United States version. Now let's go to Bing Ai and have a look at how our prompt to look like. So I opened bing.com, chose balanced conversation style. And I am pasting my prompt. And the last 200 reviews of Ford Motor Company. You analyze them and find common patterns and summarize and pros and cons. Table mentioned how many consumer reviews featured each argument. Use my source. So previously I have Googled and chose the source that has more reviews of diverse cars. It takes a bit longer for AI to analyze this information because it actually goes to the website and read this information. In this answer, I can see that has confirmed that it analyzed hundreds of use from my source and provided the summary. And each time I repeat this, I get more or less the same summary. I mean, the wording might be different, but it's more or less the same. So let's now use this summary and go to create our swat analysis. I actually did it before. So the answer was a bit different and you can see the difference. Create a swat analysis for Ford. I'll provide you with a consumer review summary. You'll create a two-by-two table. For the analysis. You use both PESTEL analysis above competitor data and consumer reviews and consumer review summary semicolon. And I'm pasting this summary, removing everything that's not needed. And hitting Enter. Here is our swat analysis for Ford based on provided information. And I can see that it has classified this information into a table and used both my PESTEL analysis before competitor data and consumer reviews. Looks awesome. I also can see that opportunities and threats are the macro factors. Mostly. It's important to not include our own ideas, but really use the external factors that we cannot really influence. Alright, so you can iterate these two steps, both for your company and your competitors company. You can even go and creates what analysis for your company in relation to your competitor. I will leave the prompts. You can copy and paste it. Now, let's talk a bit about taus analysis. Basically, it's the expanded swat analysis, but it also gives us a few ideas. Here. You can see how to use strength to work with threats. And how to use strength to work with opportunities, e.g. strength to opportunity strategy is S 0 strategies. And you can see for ideas here then strength to threat strategies. And you have a list here, then you have weaknesses. Opportunities strategies, minimize weaknesses by taking advantage of the opportunities and weakness to threats strategies, how to minimize weaknesses and avoid threats. Alright, so let's summarize what we have done in this competitor analysis and research lecture. We've learned how to collect the basic information and competitors. We've learned how to create bras and cons analysis or some college advantages and disadvantages analysis of competitors based on customer reviews. We've learned how to create what analysis and expand strategy is based on this. What analysis with tos analysis. Hope this helps you to create your competitor research and analysis faster and come up with better ideas. And as a result, when more See you in the next lecture. 14. CHATGPT FOR MARKETING STRATEGY: Strategic Positioning for Audience Segments: Now we know how to create and analyze marketing landscape. Create and segment our target audiences, research and analyze our competitors. What do we do next? Do we start writing posts? Not yet, but we actually start creating now, not just analyzing, but really created. The first thing we want to create is our positioning in the market. We have all these different people. We have five different target audience segments. So how do we create a post for all of them? Well, the answer is, we don't, we don't create one post for all of them. On contrary, we create a positioning for each of the segments. And for that, the perfect case here is for Kotler's STP. Kotler's STP stands for segmenting, targeting, positioning. The thing here is, we already have quite a large thread here with Judge OPT. And tragedy has a short memory, so it tends to forget what we spoke about previously. When the chat gets longer, Kotler's SDP fits perfect here because we already have the segmenting and we have positioning, which is actually going to be new here. Let's move on to my screen and have a look at how it works. So, so far, we have so many things here. We have marketing landscape, PESTEL analysis. We have expanded audience segments. We have Kotler's segmentation, we have user persona, we have list of competitors. So we're back to forming our positioning. Godless STP is a great fit here because it reminds judge pity on what we've discussed previously. We have our segments here. So Kotler's SDP for following segments discussed previously in the chat. And I also want to mention those segments in order to make sure we don't spend time generating response that referred to wrong segments or judge beauty forgets. So I specify those segments. Simply copy and paste them from our regular table. And I also mentioned to use table format because I want this in quite a brief format, brief overview. So we have segments targeting strategy and positioning strategy. We have all of our groups here. And the targeting strategy, charge epitope refers to different targeting strategies according to compost SDP method. If it's something new for you, don't forget to check out the supplemental materials in this section. You will find it out. It's very useful. And I'm finally, we have positioning strategy. Let's double check some of those family **** is drivers charge it, but he suggests that we position ourselves as affordable and reliable cars that cater to families needs. In my opinion. Well, that is not wrong. However, we want to specify some things. Already know some of the particular benefits of our brand. You might want to specify that. Don't forget about refining all the answers that chat GPD provides. But so far, these answers might seem a bit superficial, but they are not wrong. They just need to be verified by human professional with critical thinking, these positioning strategies are quite obvious, but we will use them to create the right content. Using this strategy, we will use these positioning strategies to create content plans and tactics. Let's quickly summarize this short lecture. In the last few minutes, you have learned what is Kotler's segmenting, targeting positioning methodology? How to use it in conjunction with Chad deputy. In the next lecture, we will move on with Kotler's methodology and create marketing mix using the four Ps. See you there. 15. CHATGPT FOR MARKETING STRATEGY: Create a Marketing Mix with ChatGPT: Alright, now that we have all the audience segments, we have the positioning for each segment. What do we do next? Need to start creating something, right? And the first thing we create in marketing strategy is marketing mix. This term was introduced long before Philip Kotler, and it involved 12th factors. And it didn't live up into Morton corporations and enterprises simply because it's not usable, It's not very actionable. And normally when we talk about creating marketing mix, we talked about four Ps, product, price, place, and promotion. As usual, let's switch to my screen and have a look how it works. So we're back to my screen. We can see our segments targeting strategy and positioning strategy. Let's assume that we confirmed this positioning strategy and we move on with it. More product insights. Then you would want to refine it and ask to GPT to add some information and rewrite this table. But we will move on. And we say, based on positioning above, create a marketing mix. I indicate that I want Kotler's methodology. Don't think that I'm just a huge fan and Ambassador of Kotler. It just makes sense to follow one methodology throughout this course to create a consistent workflow for our purpose. I indicate the brand, I want to create the marketing mix four, and then I indicate that I want a separate mix for each target audience segment. It's quite important because we cannot create one mix for all target audiences because the price is going to be different. The place where you sell the product can be different. And the types of promotion you will use is going to be different for each segment. So I just specify this. If I don't, judge, a PT will respond with one pack of four piece for all segments. And that just doesn't really work. Especially when you are huge enterprise like Ford, of course, for some small businesses, where you have one target audience segment, that might work. But in most cases, it's just not enough. I also want to keep using table format simply because it's more efficient for each word spent by judges OPT. And I also define the format of the table that I want. I want to first have the target audience segment. Then or four P's one-by-one. Let's have a look at what charge it suggests. Here's an interesting drug and that's functionality of tragedy that I wanted to show you. But I was looking for a good place to do that. And this one is perfect. Sometimes it might not be completely satisfied with the result that charge APT provides. But don't let that disappointed. Sometimes it's enough to simply click regenerate response. And you'll see that chat jeopardy doesn't come up with the same thing twice. So it's actually a way to generate more ideas. And what's beautiful about the interface. But this minimalistic interface is that all the responses are saved and you can simply navigate them. They work as a tree. If you develop more prompts after. For this purpose, it helps me to generate multiple tables like this. And I can see which ideas I like more and which ideas I find more relevant to my plans and my vision. It's important to keep your vision in mind. Talking about the plants. Unlike in pure marketing analytics where you simply analyze facts, marketing planning, you need to match the plants, match the ideas to your vision. So at this stage, I really recommend to regenerate response a few times until you see the ideas that you really like that really drive you and wants to feel that, wow, this thing reads my mind. That's when you have a match for your vision. So let's dive deeper into the responses that charging party provided here and have a look at the difference between the two responses. Let's have a look at the first one. The first one, I can see that tragedy provided type of product for my business. And in some cases that might be useful. I can see quite abstract pricing. But in the case with charging party that doesn't really do maths, That's a good point. I also want to outline that it's good when it comes to place and promotion. In some of the answers, I like promotion more. But really that might be an idea for you too. Gather a few of the best ideas from each table and creating your own. Let's have a look here and discuss this example. For a family forecast drivers. They suggest advertising campaigns emphasizing safety, durability, and reliability for families. Sounds like if at for Volvo. It suggests sponsorship of family oriented events and partnerships with family-oriented brands that make sense. I would even extend this to family oriented bloggers, vloggers, and stuff. So whoever is positioning themselves as parents, as it fit here, social media content that showcases families and in John Ford cars makes sense. I would even expand that to my taste. I would use famous families. Now, let's have a look at the second option. So I just clicked regenerate response here. I can see product examples by Ford. So I can see car models. They do make sense in most of the time. I do like the placement. They also think that for families in the United States, financing options is a good promotion plan. But what if your market is different? Well, regenerating the response and adding your ideas, refining the results. So now let's summarize. In the previous lecture, we learned how to create positioning for each segment of your target audience that we already defined with Judge apathy. And in this lecture, we expanded our knowledge about the positioning and creating a marketing mix plan. For each segment. In the next lecture, we'll build a system of our digital channels and what we communicate in each of them. See you there. 16. CHATGPT FOR MARKETING STRATEGY: Crafting a Comprehensive Content Mix with AI: Hey, what's up, everybody? So now that we have the marketing mix, positioning, target audience segments, competitors, landscape, we need to start creating some content ideas and rubrics. We need to somehow structure what we do on each social media channel. And in this lecture, we will discuss a prompt and the approach that I'm particularly proud of for me, it's definitely five-star. Let's deconstruct this prompt. You will be able to copy it for your case and just substitute the variables or create your own template that works for you. Simply change the columns. So let's deconstruct it. Create a social media content mix table. So the keyword here is content to mix. It's like marketing mix. Just for content. I remind of the defined target audience segments, marketing landscape, competitor analysis, and brand positioning. It's important because I want to match everything that I've done before. And also you will be able to use it in a new chat if you already have all of this information. I define the format that I need. The format that I came up with for this particular course. And I think that many will be able to leverage it is have the social media channel, have the purpose for this social media channel. Rubrics for each one. I would say like for rubrics would be optimal. I would want targeted audience segments for each channel. And the key metrics to track. This way. If you are, let's say onboarding on a new job or a project, that's a fast way for you to start. It's not even a starting point. It's already a good draft. Now let's have a look at the following instructions for each social media channel. One challenge apathy to identify the purpose and target audience, segment. And list for rubrics that align with Brent positioning and are relevant to the target audience segment. Use information from your defined audience segments, marketing, landscape, and competitor analysis to inform your choices. Well, you might think that I'm repeating in this prompt, but actually these are a bit of a different things. So here we have the first initial task description, then we have the form, and then we have more detailed instructions. And it's okay if you repeat sometimes because chat GPT will understand the structure of your request. Now, I outlined the social media channels that I've chosen for myself. I do this because it's actually a part of the research that we didn't do with Judge. A pity because you need to do it with your hands. And in each industry that would vary a bit. You need to find out what are the highest priority channels for your competitors in your industry. So here is just an idea of which channels and how you can choose them. But really, you need to think of what are the most important channels in this industry. Then I remind Chad GPT with target audience segments, a simply copy and paste the name of each segment. Then asked LGBT degenerate rubrics and purpose in relation to these audiences. Remember this phrase in relation? Yes, it's back here, but just in a different context. Literally asks Chad to put t to use its context based functionality. And let's have a look at why it has come up with. As you can see, I've generated three ideas, three options, and I still recommend you if, if this doesn't fit your vision, try again, try to regenerate this response a few times and you will see which option is best for you, or you want to make a combination of a few options, you will be able to do this in an Excel spreadsheet later on when you paste this table, I can see that all the columns are correct. So Facebook, the purpose is brand awareness and community engagement. Rubrics. That's interesting for its history and milestones. Alright? Safety features and family friendly design. Behind the scenes, Ford and customer stores. The target audience segment for such content would be family focused drivers and commuter, convenience seeker. The key metrics here, engagement rate, reach, and impressions. Of course, you might want to use some abbreviations and detailed metrics, but I like the idea of what kind of metrics it involves for this goal. These are all awareness and engagement metrics, which totally makes sense. Let's have a look at one more rubric. And let's think of TikTok. That's something that I'm particularly interested in here. So let's have a look at what Chad Jeopardy provides, whether it would fit in modern day TikTok. By modern day, I mean, the date of recording this. This is February 2023. So it's entertaining and educational content. Carmen, tenants and care, funny car related skits and challenges, driving tips and tricks for its commitment to sustainability. And now we're not just saving time, we're actually create things that are relevant to our audience and our brand. It's not just AI creating it for us. It's us creating it with the help of AI. And these are different approaches. I believe in this one, stronger. Let's go back to the initial prompt. What if your goals are different and you plan on using a different template, e.g. you want to include a website, blog. Make sense. What if you want to include public relations? Think of these prompts as a dedicated Canvas for your needs. Feel free to customize the moving parts, the company, the industry, the target audience segments, the columns of a table. The real beauty of this instrument is that you can customize it so your marketing problems and complete your marketing tasks with more efficiency and quality. As a result, that makes you a specialists who does more with less. And it's something that most businesses are trying to achieve right now as we're heading into a recession. So as usual, let's summarize what we've learned in this lecture. We've learned to create a content mix using our audience, marketing landscape competitors, you can customizing the cons to our needs. And you even have the readymade example of how it can be done that you can just simply copy and paste when you use supplement materials of this course. Thanks for watching and see you in the next lecture. 17. CHATGPT FOR MARKETING STRATEGY: Summary: I hope that you learned something new and added structure to your knowledge and observations that you've had previously. This tutorial on using GPD for marketing strategy. We've learned to quickly process data, create assumptions, and use marketing strategy frameworks within large language models. And you know, despite not seeing you face to face throughout this journey, I feel like we've gone through it together. So if you have any questions, ideas, suggestions. I'm here, not AI generated, and I will be happy to communicate with you and get in touch. Thank you for sticking with me and my tutorials. I really appreciate your time and attention. Keep learning, keep growing and season. 18. CHATGPT FOR CONTENT CREATION: Content is everywhere. Alright. Welcome to this practical session on enhancing your content creation skills, whether for work or your personal brand. The quest for new and engaging content ideas is a common challenge. And sometimes knowing what to say just isn't enough. Once you know what to say, which is actually a good thing to search ideas, but the real task is finding ideas and knowing how to communicate them effectively. For me, when I'm short on time, these ideas just don't flow as naturally as I would want to. And that's where Genio Chagpt comes into play. We'll look at the practical ways to use Chagpt for generating content ideas, the content itself, and bringing it all to life. We'll cover idea generation, content repurposing, copywriting, editing, and much more. But remember, it's just a starting point. The real value comes from how you adapt and refine these ideas to fit your unique vision of what's good and what's bad. And speaking of vision, I see this series of lessons as a method to inspire your creativity with the help of hatipt. So get ready to challenge these ideas. Pick the ones that you like, ditch the ones that you don't, Twick ideas that you like, at least partially, and test them against your knowledge and experience. So let's get started. 19. 14. Introduction: AI for Creative Idea Generation: Hey, what's up, everybody? I'm so happy that you're here. And actually, I want to congratulate you because you've gone through the hardest part of the course that also involved a lot of marketing strategy. And generally, strategy is considered to be the hardest part of marketing. But why did we start with something that's more complex? Logic behind this course can be described in one sentence. Start as a strategist and finish as an artist. Now that we have the strategy, why would we involve AI algorithms and creative work in something that's humane? Does it take the personality out of the work? Doesn't describe it. The author, the creator. Well, not necessarily. Let me quickly go. A few years before Chad deputy, everyone tended to think that AI is going to substitute some very generic, repetitive tasks and it's going to fail at creative ones. But then we saw mud journey that created images with AI than we saw Delhi. That also creates images than artists started creating music with AI. Looks like it has gone in absolutely different direction. And it's actually not such a surprise because there are techniques and practices and generating ideas. Alright, let's not make introductions too long and jump right in to generating ideas for content. 20. 15. Generating Content Ideas with ChatGPT: In the next couple of minutes, we will learn a way to generate hundreds of ideas. Those ideas will be tailored for a particular channel and will be quite specific to what you indicate. As usual, let's go through a quick step-by-step guide. So step number one is to assign a row of a social media marketer. We also want to create a particular amount of ideas. For this lecture, we'll try ten. Want to indicate the channel, the company. If the company isn't known to check jeopardy at the moment, you can describe your company or industry. The ideas should be related to a rubric that you have in your content mix for this particular channel. In step number two, you will look through the ideas that are still high level but all ready, relevant to what you work with. And ask Djibouti to elaborate on one of those and create ten more ideas for the ideas in the previous response. Using this algorithm, you can create 100 of ideas from just one batch of high level ideas. And all of those are going to be more specific. Then what charge? But it gives you the first sounds exciting. Well that's not it. You can iterate on all levels. So level one is what we get after step number one. Level two is what we get after step number two, and this can go on forever. So let's have a look at more details and what charging party actually gave us when we used this approach. Here is pumped number one. Step number one, act as a social media marker, create ten content ideas for Ford Motor Company. They, they should be related to driving tips and tricks. Quite short, but still, in these two lines, we have quite a lot of information. We have the row, we have the amount of ideas that we want to generate. We have the channel, we have the company. We have the rubric, just two lines, but this prompt is very specific. And these ten ideas that are supposed to be high level, while they are high level, but already quite detailed. I would say. Now in the step number two, what I want to do is to elaborate on one of those ideas. The idea is share tips for parallel parking, backing up and maneuvering in tight spaces. I think this should be particularly interesting to commuter convenience seekers. And here we are time-lapse video of Barlow parking and type spot with helpful tips and tricks. A POV video with helpful VoiceOver on how to angle your car for a perfect parallel park. Alright, let's move on. Humorous take on the difficulties of parallel parking and how to overcome them. I mean, all of this already sounds more actionable to be. This can move on forever. Actually, the step number two, if you want to iterate on other ideas from the first stake, we can actually repeat the second prompt and just change the number. However, there's also one more idea for organization within the chat. You can simply add it your second prompt, and then you will be able to navigate in the second prompt as a carousel, very handy if you are working with many ideas in the same chat, but this one is more for demonstration purpose. So I just created a new prompt and asked to elaborate on video ideas for option number six. And we have driving in the rain. So it's all about the weather conditions here, adjusting your speed and following distance. I would even say something about the mirrors. I mean, all of these ideas are quite specific and you can start creating videos from them. The next step is you want to create this content. You'll probably need to create an outline for the video, the scenario with timing, scene description, and maybe any scripts if there are any. So that is something we will discuss in the next lecture. But first, let's summarize this one to create a hundreds of ideas with judges OPT, we first need to create a prompt where we set the task clearly. Then we need to iterate on each idea that aegypti provided in the first iteration step can continue on the second level as well. You just need to specify what you want to iterate on it. If you don't, it will iterate on the first date. So see you in the next lecture. 21. 16. Scripting TikTok-Style Video Scenarios with ChatGPT : Alright, so in the previous lecture, we have learned how to generate hundreds of ideas in a matter of minutes. Now what do we do with them? Are they enough to shoot a video for them? Well, maybe if that's your cup of tea, this course, we are here to learn what Chad Jeopardy can do for us. I think once you have the idea, you need to have some more ideas on how to execute those. In other words, in this lecture, we're learning how to create a video scenario for one of the ideas generated in the previous lecture. Let's have a look at the prompt first. The prompt is for idea number one, create a TikTok video scenario. The video should have an unexpected humorous twist in the end. Put it in a table that should include the following cons, semicolon, timing, scene name, Cindy details, scene description. Now, this is the form that's typical for contractors when they work with video. And when you have this kind of outline for your TikTok video or Instagram, real or whatever video content that you're planning. This work is going to be much more predictable and easier to prepare for. In general, it's just the way it works when you create a video. So as I know that TikTok loves this unexpected, humorous twist. And people on TikTok are there for kind of entertainment. I'm just adding this in the description and I'm specifying what kind of ingredients I want in the table as well. So let's have a look at the idea. Let's read the humorous twist first. The humorous twist is that the driver sees a unicorn in the backseat, which is unexpected and adds a touch of humor to the video. The unicorn can be used as a fun way to show it how adjusting mirrors correctly can help you see things you might not expect. Alright, well, this sounds like it makes sense. So there's timing. The tragedy suggests that it's going to take around 14 s, anything that's okay for TikTok. So intro is a general shut off Ford vehicle driving down the road. It's just an introduction. Then the driver adjust their merits and a magical sound effect is hurt. I already can imagine how it sounds on TikTok than the driver looks in the mirror and sees a unicorn sitting in the backseat. Alright? The driver takes a double take and it looks back the mirror, but the union is gone. Okay, there are some dynamic here. The driver, it looks confused and tries to figure out what just happened. The camera shows a glimpse of the unicorn hiding behind the front seat. And Ford logo is displayed at the end of the video, right? Well, that makes sense for me. But there's one thing I'm looking here at the end of the video. That is a copy line. A couple of line is a phrase that typically explains everything that was going on in the video. Usually in the video add it comes in the end. So I simply asked Judge APTT provide three copy lines for the ultra. Another charge a pretty knows what is the copy line. So there's not going to be any problem with that. I can see three options here. Let's decide which one is the best. Here are three capitalize for the ultra, stay safe on the road with Ford's advanced safety features. I think that should be that sounds good, but it's probably for a different topic. So that's a miss for this one. The second one, drive with confidence and see the unexpected with Ford. That explains the video very well. And it also matches that you should see the unexpected objects. I liked this one because it precisely explains what happens in the video. And benefit that you can see everything and experience the magic of driving with Ford. I like the thing with magic, but the problem is it doesn't direct us to any particular benefit. So I think this second option explained the video a lot better. I would then typically add this to my video in the end. Okay, so let's summarize. In this lecture, we have learned how to create a scenario for a TikTok video. This could be your YouTube video or Instagram or any other video format. See you in the next lecture where we will adopt this video to different formats and platforms. Take care. 22. 17. Content Repurposing Made Easy with GPT: Hi and welcome back. So in the previous lecture, we have learned how to generate ideas for a video script. Well, the thing is that nowadays it's no longer enough to only post one video on all channels. And it's going to work well. According to data reportable. 99.9% off social media users use more than one social media channel. And we want to give them a reason to follow multiple channels. And each channel once some unique contact, not just to copy. And each channel's algorithms tries to value unique content higher above the copied one. So now let's turn our unicorn video for TikTok into posts on other social media. And traditionally, let me invite you to my screen-sharing. Alright, so we are back at our TikTok video idea. My next prompt is very short. Depth this idea to Instagram carousel post. For this one, I can see that I could use the same idea just in a different format. Let's have a look. And I have slide by slide looks six slides here. Imagine Ford vehicle driving down the road with a caption. Proper mirror adjustment is key to safe driving. Close-up image of the mirrors with caption. Adjust your mirrors for optimal visibility and safety. And humorous image of a unicorn sitting in the backseat with the caption, but who knows what else he might see? Always check your mirrors. Before you hit the road. Experienced magic of driving with Ford's advanced safety technology. Not sure I want this one. Maybe I'll just cut it. Slide six, final image of a Ford logo with a caption dr, with confidence to see the unexpected with Ford. Alright, so I don't think this post would particularly work perfect for organic. But I can still see that people would want to see to the end, especially I think they might be captivated on slide three. So I would say boosting such carousel post might actually be not a bad idea. Now let's adapt it to Twitter post. Stay safe on the road. With forts, advanced safety features. Remember to adjust your motives for optimal visibility and see the unexpected. Who knows what else you might see and hashtags. Well, I would remove the first sentence again simply because I didn't instruct judge a beauty here that I don't like the first copy line, but now I can do it myself. I actually like the phrase, who knows what else you might see. It sounds quite intriguing and shareable. I could almost imagine people commenting with what kind of weird things they can imagine seeing on the backseat. Which could actually be a good engagement call to action as well. You see, when you have some ideas, It's easier to generate more it from your side. Let's turn it into Facebook post. And let's go back here to our table with rubrics. Here. On Facebook. I can see that we have behind the scenes as one of the rubrics. So let me go back here. And I asked to turn it into a behind the scenes post on Facebook. So I indicate what kind of posts they want. We actually align with our content mix. Let's have a look. Ever wonder how we make our afford ads. Check out this behind the scenes, look at our latest shoot where we got a little creative with our mirrors. Proper adjustment is key to save driving. And we wanted to showcase that in the font and unexpected way, wouldn't you? A unicorn would make an appearance onset. Stay tuned for the final lab, where you'll see the magic of driving with Ford's advanced safety features than the number of hashtags. I mean, this sounds like quite a human communication. Not bad at all. Right, Let's summarize these two lectures. We've learned how to generate ideas for series of posts and how to distribute one content piece into other channels. And all of this is done in a matter of few minutes. It's scalable. You can create a lot of content like this. And it looks kind of good. Honestly, I think it's better than some of the ideas I've generated myself in the best. Well, some of them were really crappy to be honest. But that's just a part of the process. Alright? I hope that with these techniques, you will be able to create more content, make it more relevant to your strategy, and you will stay creative and on your schedule. C, you in the next lecture. 24. 18. Quick workflow: Write Like Pros: In this lecture, we will learn how to get a sample of the copy that you like and create one for your bent and audience. Let's go to my screen and I will show you how. So first, let me show you one awesome research online. I'm not affiliated in any way, but I haven't been a follower of this resource for two years. And it's been my awesome, huge source of inspiration, the resources, marketing examples.com slash inspiration. Here you have a list of structured awesome at examples from different spheres. And when he hover on the screen, you see the reason why similar example of works. So as you work with such kind of inspiration, you also learn your work and learn and that's important for us to always be learning while working. So what we're gonna do, we'll take a copy, ask GPT to analyze it and create one as a pattern match. I've chosen one for you right now, and that's this one. Very vivid example. One-click, no passwords. The world's fastest checkout. How can we possibly transform it into an ad for Ford? Let me show you the prompt. So here we are with the prompt. It says, act as a creative copywriter. You remember assigning roles, then use this copy as a reference. Semicolon. One-click, no passwords. World's fastest checkout. Please keep in mind that punctuation is very important for clarity of our prompts. Analyze it and create a similar at copying in relation to my brand. You see this is in instructions, so you will do something. You can't use the same arguments. I don't want to get confused. I don't want to talk about passwords here. Use the same length and sentence structure. Now, business data company, Ford, but for Instagram audience, Venture seekers, let's have a look at the response. Here's the responsible each IGBT, original ad copy, one-click now password. So world's fastest checkout analysis. So not only you get the cup, you also get the analysis, the expanded analysis on what it is. Then we have new ad copy. One turn, no limits. World's most adventurous, right? How do you like it? I think that could make a very good headline or copy line four, the car, this relevant for this target audience segment, or a banner, like a billboard in the city. Why not? But will it work with any reference? Let's try one more. I'm a huge fan of eight reps as well. That's my favorite SEO tool with a trap. So you don't have to be an SEO pro to rank higher and get more traffic. And we can see the emoji here as well. Let's have a look at how this worked. I use exact same prompt here. Act as a creative copywriter. Everything's the same except a different example. Now, here's the original copy. And now here new ad copy with Ford. You don't have to be a car Pro to conquer the commute and save time. Alright, well, probably I would change car Pro to something more specific, e.g. a. Driving Pro or race car driver or a car mechanic, depending on what kind of objection we're trying to cover right now. For you. It's just a matter of finding the right inspiration and then adapting it to your audience. By the way, I was doing this in a new chat, so it doesn't have the whole context of our previous conversation with Judge Judy because they simply don't want to make it very cluttered, But I already know what I want to do. I already know the target audience Name the judge generated itself. So it looks quite get from me at the moment. And you can apply it to longer-form content as well. But I find that this idea works well for short copies. In particular, you can also use this with videos. Download a video, Let's say from a TikTok ads library, use any speech to text tool. It's simply a tool that would generate text from a video, paste this text as a reference, and use the same prompt. Then give it to the speaker of your video and you have a reference that's done. But the logic stays the same. It gives a reference asked to analyze and replicate. But my favorite approach here is to go to marketing examples that come. It's a free resource. It uses at examples to show the classic creative techniques. And that's why it's not as much of a stealing from one particular video, but just applying a particular technique. I hope that lectures in this section have inspired you to create. 25. 19. Reasons NOT to use GPT for copywriting: In the next couple of minutes, we will discuss why using charges beauty for copywriting isn't always a very good idea. The first thing is its accuracy. Check GPT can be very confident in whatever it says. However, is that true? In many cases, I found that it confidently claims to information that is completely not true, or at least a bit far from it, you will see the numbers, will see the sources if you ask for them. But the information ends up being a bit incorrect. Although very persuasive. Second research, this one is a bit related to accuracy. However, it's a bit different. Many people tried to do a research completely with Judge petty, but it's not always an option because it's simply does not factor check. Let's keep in mind that is language model not a fact checker. The third one is tone. If you don't do an extra step to use some particular tone of voice, change in PT will respond in a very typical way. And once you start working with it a lot, you will see this tone, like from a mile away, you will see some typical sentence constructions and paragraphs. So the tone always needs a special attention from you. The fourth one is plagiarism. The problem with aegypti is that it doesn't claim to always generate unique content. So always check your content created with IGBT, with anti plagiarism tool. Fifth one is long form content. Basically, I'm not a fan of using GPT for long-form content, especially the long form content is typically quite generic and easily detected. Although at the moment, Google says that it doesn't care whether the content, what's created with AI or just with a human, as long as the content is useful to the end-user, Google is not the only platform where you distribute the content. And also this claim could change with time. And you might want to create content that lives for a more extended period of time. So let's summarize the five factors that make charge of t not a very good idea for copywriting, accuracy, research, tone, plagiarism, and long-form content. Remember a set five-and-a-half reasons. The thing is, if you work on some projects for an extended period of time, you know, so much context about it that he can never feed so much into an AI tool like judge, if T, at least at this stage of its development, at the moment, you know the emotions that you feel about the product, the culture of your particular company, the feelings that you're consumers have about your product. When you write a copy, all of this blends in and you create something more humane. Of course, human copy is not perfect, and that's what makes it beautiful. So never forget about this part of copyrighted, but don't get disappointed about GPT yet, because in the next lecture we will discuss the advantages of using it for copywriting. And believe me, they're awesome. 26. 20. Reasons to Use GPT for Copywriting: Alright, welcome back. This lecture is going to be way more positive than the previous one. Because we are discussing the advantages of using adjective for copywriting, there are way more than five-and-a-half reasons. Let's start with them first. Generate ideas we already discussed in the previous section. So let's not waste time here. Second, create outlines. You can get charged a potato. Create an awesome outline for your long-form content or your video. Typically, planning takes a lot of time. This is where we charge a bit. You can actually save a lot of your time preparing that content. Three, create ads and ad variants tend to putty nose some techniques of ad creation. You can replicate it or create variations or create test options. We will discuss it later on in the course. It can proofreader, copy, and provide feedback for it. And it's a huge thing here. Charging party knows all the copywriting frameworks and can apply them to your copy. Number six is using artistic devices to deputy knows all of them and can generate metaphors, are parallelisms or analogies that you can spend a lot of time thinking of. This can become useful when you are writing something yourself and need just this extra bit. E.g. at the beginning of this course, I was using metaphor with tragic as a chef. Well, that was actually a metaphor created with charges. Of course, as you can hear, my speech doesn't really sound a lot like judge, a pithy, right? But I use that metaphor when I was trying to fill the gap up how to explain it to someone who has never worked with Chuck GPT or technology. And I thought connecting it to a concept that anyone can understand would make sense. That advantage number 77 is you can give your content any form you need one to transform something into a short for great, one to expand. Awesome. Want to have some opposite opinion to what you have just written. Great, That's what you can do. One to create a video from a post. Want to create a post from a video? That's what you can do with Judge Judy easily. Idea number 88 is you can automate your repetitive copywriting tasks, like writing meta-description speech titles, or product descriptions. Some copywriting desk don't require a lot of creativity. Well, that's where Judge Judy can help as well. One of my favorite advantages is this one, the last one. And it's number nine. Get feedback. It can give you a copy and asked to provide feedback for it. There are many use cases when you might want to have a feedback, e.g. you're planning an A B testing and you want to GPT-2, explain when this A B test would work and what's particularly interesting about each variant, you may want to have a comment on your style and tone, and you may simply want to improve as a copywriter. Alright, so let's quickly recap. The reasons why using chat jeopardy as a copywriting tool are the following. You can generate ideas, create outlines, create as an ad variants, proofread your copy is all copywriting frameworks brainstorm using artistic devices. Give your content the right form, automate your repetitive copywriting tasks and get feedback. Nine awesome reasons why using charge if T is a good idea for copywriting. They're way, way more of these ideas. But let's first uncovered these ones. See you in the next lectures. 27. 21. 10 Powerful ChatGPT Copywriting Use Cases [WITH GUIDES]: Alright, what's up, everybody? This lecture is going to be faster because we will discuss ten copywriting use cases. So by this moment, we've done a lot of prompting for marketing in charge of beauty. And I expect that you started understanding some principles that you can apply and adjust to your marketing needs and your marketing use cases. But I wanted to show you Tan copywriting use cases that actually work well with charge deputy and that use chat GPT strengths for copywriting and avoid weaknesses. In this lecture, we'll go through the use cases one by one very quickly. I'll show you the prompt and I'll show you the example. But we will move really quickly and I'm sure that you can already use all of this information. But if you still have questions, welcome to Q&A section of this course and feel free to simply ask what you're looking for. That would help me create this course better with every new update. Let's switch to my slides and have a look at the use cases. Use case number one is create simple outlines for SEO articles. Actually in this video we are going to have to alkaline types. But in this case the prompt sounds like this. Create an essay or article outline optimized for these keywords. I suggest that we research keywords with tools that actually scrape the search. Tools like SEM rush, a traps as the ranking or at least Google Ads Keyword Planner. This way, you can get actual information on keyword difficulty and search volume. Unfortunately, as Egypt, it isn't connected to internet, isn't groveling. Web data regularly, cannot give this information. So please do your keyword research in a tool that's designed for it. But once you have your keywords, charge empathy can help you optimize everything you do for those keywords. This case, it's an outline. Feel free to pause this slide and have a look at what we have as an example. But so far, Let's move on to a more expanded outline. This one makes use of an ECE framework by McKinsey. So it says using MVC framework, create a detailed website, blog, article, outline optimized for keywords. List your keywords included a list of semantically similar FAQs using the vector representation technique using Markdown format. Please don't write the article on the outline for SEO writers. You can remind me what I asked for. You can apologize, you can self-reference. Actually, this is one that I've slightly borrowed from AI PRM app. In the end of the course, you will have the reference for this one and a slightly modified it to fit our needs and the way it works at the moment. But I really like this one. I will link it in the resources. So here is how this outline would look like. If you can create an article that solves customers problems like this, I'm quite sure it will succeed. Ranking some backlinks wouldn't hurt either. Then product slash feature name. This one is actually something that OpenAI recommends themselves. And I took this one from Open AI's library. Yes, there is a library of examples of prompting. But to use this one, we need to go to Open AI playground, Jews, modal texts, Da Vinci 003 and set temperature to 0.8 or one. Just for you to know, temperature is the degree of randomness of word choice by Jeopardy. Basically, the higher the temperature, the more creative the copy is going to be, the lower the temperature, the more precise and predictable the copy will be in the playground, right? This prompt, basically this prompt serves a better example. And then exact same patterns, but your information. This gives amazing product name ideas. Try this one out and actually run through the library at open AI playground. Link, the link in the resources of this lecture as well. It can teach a lot more for what you can do with open the eyes playground. But let's move on so far. Number four is subject lines for emails and headlines. Here, I'm asking Chad deputy, to generate multiple types of headlines or subject lines based on the copy that I provide. E.g. let's think of an email subject line. So we would write 25 email subject lines. Five, catchy wants, five creative, five persuasive, five intriguing, and five click. But the topic is Corrigan, meet up in Paris, organized by the pad suppliers shop. Length up to 50 characters, which is pretty much standard for an email subject line. Let's go to the GPT. I'm going to choose JPEG for. And you can see that it has grouped all the titles in sections. This way, you have a better choice of what you want to come up with. Number five, writing Meta descriptions and page titles. Instead of writing Meta descriptions and page titles on your own or copying and pasting things into tragic beauty for each page separately, we'll do the following thing. We'll create a spreadsheet with URLs and keywords. So number one is for URLs and column number two is for keywords. Paste your URL and URL and type the keywords that you want to target with each page. And the Meta descriptions then go to Bing I chat because we actually want internet connection for this task. Then we type, put this into a table and paste our table straight from the spreadsheet or XO. The next step, we asked to expand this table and create SEL Meta descriptions and page titles for each item. The Meta descriptions and page title needs to be optimized for keywords and come to copy the results and paste them back into your spreadsheet. Now you have four cones, URL keywords, page titles, and Meta descriptions. Five simple steps, less than 5 min. And you're done with meta-description sand page titles that are actually optimized for your keywords. Whether you have three links or you have 300 links, this is going to be exactly the same. Awesome, right? Copywriting, use case number six, proofreading. We're all human and we make mistakes. Can take this human factor out of the equation. So here is the prompt that I'm using, proofread my copy as a professional marketing copy editor, use US English. Well, that's because I'm using US English in my copy most of the time, but feel free to modify it to your taste. And then I paste my copy, my copy, semicolon pasting my copy as an optional refinement. Feel free to ask, charge a bit you to clean up your copy, saying clean up the copy and remove excessive vocabulary. I recommend to use this prompt as a secondary one as a refinement. Let's move on. Copywriting use case number seven, AB test copy. It's an excellent use when you already have a copy for your ad. But you want to test something. In this case, we are asking the following. You're a seasoned PPC ads specialist, create A B test copy. I'll provide you with an a copy variant. You'll create a copy variant and suggest a hypothesis why it could work better. Then you paste your copy, a title, and body. Feel free to adjust this a little bit to your needs, e.g. if it's Google ads or Instagram ads or Facebook ads, if it's a search ad, specify this information and judge a beauty will provide you with a better information. Also feel free to emphasize on some benefit or feature, but don't forget to test one variable at a time. Let's move on. Copywriting use case number eight, right? Or rewrite your copy using copywriting frameworks are formulas. Gpt is a language model that knows the copywriting formulas. So why don't we use it? It's especially useful for conversion, copyrighting. The brown sounds like this is the format, copywriting format to create your type of content about your top, the audience would be your audience. Use these arguments or emphasize these arguments. Here are some of the copywriting formulas that I've tested and that worked well for me. I either PAS, BAB six plus one for peace, What's in it for me, and inverted triangle, awesome stuff to play around with. Also feel free to refund your copies that you already have using these copywriting formula. Don't forget that when you're writing with charge equity is sometimes need to regenerate response simply to get the option that you like more. Don't worry, all the previous responses are saved and you can navigate as if it's say, carousel, very convenient. Charge repeated copywriting. Use case number nine, great ad copy for your landing page. Option number one, go-to being chat. Ask to create five Google search ad copies for your landing page and paste your landing page link, specify your audience, maybe some keywords. I also like how it puts the responses in a table format. That's why I'm asking the user table format with the following columns, title, body, feel free to customize it to your audience. E.g. you might want to add a keyword if you don't have been and you want to use charge a pity because it's more convenient to you. You can also do it, especially if you're a landing page is not really long to do this. Our first step is to copy the text of your landing page. Go to charge a putty, type in, put it into a table based your copy and hit Enter. Once you have your landing page in a table, use the same prompt. Hasn't been, just don't add any links. Copywriting use case number ten. It's last, but not the least. Actually, this is something that I recommend doing most of the time. The thing is charged. Gpt doesn't have a slightest idea what kind of length you want chat GPT-2, right? So why don't we specify the length that we want? In most copywriting France, you can simply add length, semicolon, and amount of words. Also, don't forget that you can use this as a refining prompt. E.g. rewrite this copy, length, semicolon the amount of words. But if you need an exact, really exact amount of words, double-check the final copy with a word counter. I noticed that in some copies, charge of beauty can fluctuate around like five to seven words. Not really critical, and can be fixed menu. Alright, so that's it. We have awesome copywriting use cases with Chad GPT. Let's summarize. We can create simple outlines. We can create extended outlines. We can do product and feature naming. We can write subject lines for emails, or we can write headlines. We can write bulks of Meta descriptions and page titles. We can use chair deputy for proofreading, writing, AB, test copy. Use copywriting frameworks, create ads for your lending page, and customize the length of your God. I hope that these ten tips save you a lot of time working with marketing, copywriting. And I also hope that they can inspire you to integrate them with your other use cases that you come up with yourself. Thank you and see you in the next lecture. 28. 22. Ensuring Brand Tone Consistency with ChatGPT: Hi and welcome back to the course. In this lecture, we are talking about how to bring our unique style and how to match our brand tone of voice. How do we avoid sounding exactly the same? Everyone else does with Judge of beauty. Of course, there are a few problems with matching brand tone of voice with Judge beauty, but there are also a couple of opportunities that are not on the surface. So let's briefly remind us, what is the tone of voice? Tone of voice defines what your brand communicates and how it does it. You can think of this as a tone standing for what your brand communicates, and voice for how your brand communicates. This lecture, we're going to refer to Nielsen Norman Group. Quite often it's a super reputable resource. I would even say that it's one of the top resources and UX and consulting world. They have conducted a study to find out how consistent tone of voice impacts users. I'm leaving a link here so that you can check out the study. But long story short, the impact is huge. The trust, the desirability, and the net promoter score are influenced significantly if your brand uses a consistent tone of voice. Let's now discuss some problems of matching tone of voice with Judge OPT first, usually brand tone of voice documents are lengthy and are definitely bigger than what charge of beauty can process. Second, tone of voice documents are often at least partially subjective and very emotional. They are integrated into corporate cultures and therefore charge if it can't always understand them the way you do the first and quite obvious idea to give tragic but T, a piece of a copy written in brand's tone. And then asking to sample it and write in the same tone as many people on YouTube say, actually doesn't really work. I have tested it using Nielsen Norman groups, samples for tone of voice. And it really didn't work. It didn't define them correctly. And when I tried it for the brand I work with, I also saw that the tone isn't actually matched. Extract that were the future more advanced models of Djibouti. This is going to work better, but so far, that's what we have and that's what you have to work with, but no worries. There are some solutions and workarounds and we're back to Nielsen Norman Group. So how do we actually match tone of voice? Well, first, there's not gonna be a 100-percent match without your involvement and personal editor, but we can get close. Let's get back to Nielsen Norman Group study on tone of voice. They have researched and identified for four dimensions of tone of voice. So it's either funny or serious, formal, casual, respectful, or irreverence, enthusiastic. And matter of fact, I'm leaving you a link so that you can see more samples of this text and read this study. Highly recommend this one while using these four dimensions, can get us quite closer to what we need for tone of voice, that's not completely enough. So if you look through the Nielsen Norman Group tone of voice study, you will see that they have identified 37 tone descriptors. I went further and classified them for you in those four dimensions. As you can see right now on the screen, you'll have four dimensions and you have tone of voice descriptors. As you might be guessing, we are going to use those descriptors in our prompt. Good news is quite good at understanding these ones. But here's one more important thing. Let's talk about mixing those descriptors. If you don't have a ton of voice defined yet, my advice would be to keep this approach of dimensions. Let me explain in more detail so we don't try to be both fun and series at the same time. We pick one side of the dimension and then we pick one more dimension. This way, you'll pick up to four descriptors. But after testing all of this with tragic Putty, I can recommend using descriptors from different dimensions. So now let me share with you the front template that works for me for quite a lot of short to mid length copy. The first part of the prompt is write a type of content. Explain what it is about where things are happen and when things are happening. So you can actually include those W5, like what, where, when, and why, then you add tone of voice dimensions or descriptors. I recommend descriptors, but dimensions also work well. I recommend to specify the length in words. Most of the times when you are creating a copy, he know which length you're expecting. You can always imagine it further. You know the placement. And I recommend to specify this one here. And of course, describe your audience. But using those stone descriptors isn't the only way to match tone of voice, especially if you already have your tone of voice defined in your brand, you must likely have such point as brand personality. Most own voices describe their brands as a person and this is a term that child to PT also understands. The second option for a prompt template. Be exactly the same except instead of tone of voice, you would say brand personality tone. Then describe your brand as a personality, e.g. that could be sarcastic and nerdy geek or serious child, or a superstar, or whatever you can think of when it comes to your brand, just make sure that it resonates with your target audience. Now let's have a look at a few examples where I had exactly the same prompt except I tried different tones. Let's imagine we're writing an announcement for a Carnegie meetup that is happening in New York, Central Park on March 20. Let's try the first tone of voice. It's going to be funny, casual, irreverent, and enthusiastic. As you can see, they are from different dimensions. The length is 50 words, audience, female, married, no kids. Again, this is just a hypothetic copy. The audience for this meetup could actually be a lot different, but let's just keep it for the sake of educational purpose. So I can see here that is actually trying to be quite playful. I like the start where it calls out the audience. I can also see these font attempts like for real tail wagging time, barking, good time. I think that's quite playful. I mean, tragedy is using wordplay here and is doing quite well. Let's have a look at another example where we made the same request is different. Don't write an announcement of a corrugated up that will happen in New York, Central Park on March 20th. Tone of voice, romantic but also sarcastic. So you can see I'm trying to create a contrast with the tongue descriptors from my experience, these contrasts work quite well in terms of voice. Ladies, grab your Carnegies and join us for a heart fluttering, tail wagging. Meet up in New York's Central Park on March 20. Bring your best romantically and let your pups run wild. Who needs a man when you have a quirky, see you there? Or should we say forever alone according left, no husband, no problem. Alright, I would better and not by a query. After my wife reads this. Alright, anyway, for me it has achieved the descriptors if you want to create a few more attempts, so you can simply hit regenerate response and you will see a few more attempts. No worries. Your results are never erased and charge a PT, you can simply scroll left and right. When you hover your mouse over the logo, the average rate of change. Let's have a look at one more. Everything's the same as just carrying but informative tone of voice, attention. All lovers join us for a fun-filled day at core gimme top in New York's Central Park on March 20th. This is a great opportunity for your furry friend to socialize and make new friends. Don't forget to bring water snacks, and **** bags. See you there. Well, this Copy Features couple of advice. So there's some information on what we're, and then it's also kept here in the first part of the message, but also There's some care about both the dogs and their owners. There's also a tip to take water, snacks, and put packs alone. And it's kinda more informative and carrier. So as you can see, these things actually work. Alright, let's summarize. Matching. Tone of voice isn't an easy task both width charge deputy and without judge. But Nielsen Norman Group study can help us do it with charge epidemic. We have for tone of voice dimensions, funny versus Serious, formal versus casual, respectful versus irreverence, enthusiastic versus a matter of fact. Even further descriptors for these tone of voice. They mentioned implementing those and our prompts is going to let us be closer to our brand's unique voice. Hope this safety a lot of time, but don't forget to let your content go through your human check. Don't get too carried away when writing with Judge if Team see you in the next lecture. 29. 23. Incorporating Stylistic Devices for Creative and Engaging Copywriting with GPT-4: Remember in the beginning of the course, we discussed starting as a strategist and finish as an artist. So in this lecture, we're covering the part of the artist. And actually we're learning how to make our copy more creative, more unique, and stand out from tons of AI generated content. And to do that, we'll use stylistic devices. They can also be called creative techniques, but basically these are the same tools to accomplish that. Our first thing to do is to choose the right stylistic devices. There were tons of them. Most of them come from the classic literature. Let's categorize them and discuss which ones work better with Judge APT and which ones work worse and how to actually integrate them in your copy. So when we're talking about eight categories, sound and structure devices like alliteration, assonance, consonants, parallelism, and chaos. Miss figurative language, allegory, metaphor, simile by signification, metonymy, synecdoche, key emphasis and engagement, hyperbole, irony, rhetorical questions, repetitions like anaphora and before, contrast and comparison, antithesis, wrong, paradox, reference and delusion, brevity and clarity, question-and-answer, arrangement and progression. After I said all of these, so far, up to version Djibouti for the best and easiest, wants to use our alliteration, assonance, consonants, onomatopoeia, oxymoron, parallelism, and miasmas. So basically these are just two categories of stylistic devices, sound instructure, contrast and comparison. So here's the prompt that works for me. Generate a product description. Actually that could be a different type of coffee, whatever you're writing for a product or company with the focus on your device. And then type stylistic device. It's important to add this stylistic device in the end so that we can mark what we actually mean. We don't want consonants to be messed up with a feature in the prompt. So we want to specify that it's a stylistic device. Alternatively, you can say, use this stylistic device, semicolon, then name your stylistic device. I tested both of these options and they worked equally great. However, I must mention that sometimes it takes two or three regenerate response clicks to achieve the result that you really like. So just give a few chances, more, don't get disappointed right away. Also, I suggest that we mentioned the audience, the tone, and the length. I assume that by this moment, if already watched the lectures which explain these concepts. And you can layer all of this knowledge together to achieve the results that you want for the other groups of stylistic devices, it's harder for each IGBT because it requires more contexts. Can it be done? We just need to do one little extra step. First, create a copy that you really like. You can use any of the previous tips for copywriting. So anyway, we just need to come up with a copy that we like. Or a copy of that at least conveys the message that we want to convey. Then in the second prompt, we need to refine or modify the copy. And that prompt would sound like rewrite using the stylistic device and name your stylistic device. Feel free to use any other parameters like length if you want. There's one more cool thing that I wanted to show you before. We have a look at all of the examples of these steps. This one is my favorite. It's actually style that I looked up when I read 0 Henry's short stores, I found out that he often uses sentence rhythm in the paragraph. And this sentence with them starts with a short sentence, followed by two more compound, longer sentences, and then ending the paragraph with another short sentence. I really liked this style and I really like using it in marketing, copywriting. And I thought this one might be cool to share with you. The prompt sounds like this. Write a paragraph about the topic and the following sentence rhythm. One short sentence to compound sentences. One short sentence. For me, this sounds very readable and memorial. And believe me, if you do this, most of the other marketers who are playing around with IGBT are not going to use it, which make you stand out from the ground. Now let's have a look at the examples. Here is the prompt and the only variable changing here is the stylistic device emphasis. Here it comes, generate a product description for a fitness studio. Stylistic device emphasis, assonance. Assonance means repeating vowel sounds and words that don't Ryan. It's not repeating in every word, but it's building a pattern within a sentence. Let's have a look how it worked. Welcome to our fitness studio. You can see this IOU, IOU pattern here where your ambitions come to fruition. Not really sure what this word means in English, but we could change it ourselves, right? Our classes cater the young and daring, designed to boost endurance, balance, and strength. You can see the patterns repeating here as well. Our instructors are passionate and experienced, and our community is supportive and fun. You can hear this repetitive vowel sound in the middle of a word. Come join us and embrace your fitness journey today. And it's not obvious, it's not in every sentence, it's, it doesn't sound very artificial, but it does give a melody, does give us sound to your copy. Now let's have a look at consonants. Everything's the same except the creative technique is consonants. Introducing fit fusion, the ultimate fitness studio for young and ambitious go-getters. Our classes combine cardio, strength and flexibility training for a full body workout. The leave you feeling fantastic. From KKBOX into yoga are expert instructors will guide you through each class with contagious energy and enthusiasm. Come join the fitness fusion family. Today. In this short paragraph, you can see three uses of this one. It's classes combined, cardio, feeling fantastic and fit, fuse and family. That's a check for me. The next creative technique is onomatopoeia. That's basically sound following. So it's referring to sounds produced in nature. Welcome to fit. Boom, the fitness studio that will make your workouts go. Boom. With our high-intensity classes, you'll feel the sweat sizzle and your muscles pop as you power through each repetition. Our trainers will push you to your limits with motivating cues that will make your heart go thumb. Come join the fit fam and experienced the fitness explosion at food boom. I can see the users here and they are quite elegant. I mean, there are not too obtrusive. They don't interrupt the copy. For me, it's still usable one, Let's have a look at oxymoron. It's the first in the series that we have looked at, the one that's about comparison and contrast basically combines the opposite concepts. Let's have a look. Welcome to our fitness studio where we combine intense workout with a relaxing atmosphere. You see that's exactly what oxymoron is doing. Combining intense workouts with a relaxing atmosphere. It's also combining opposite concepts. Our trainers will push it here limits while you unwind and our serene environment. Join our community of motivated individuals and experience the amount of sweat and serenity. Let's get fit and then, well, actually that's one thing that I noticed here is that sometimes when you prompt to use a certain stylistic device, tragically t finds it. Okay to use the name of this stylistic device within a copy. I don't think it's really okay. So I would just then go and remove this word out of the company. So here I would simply rewrite the part of the sentence of experienced the similar enough sweat and serenity. You can simply say experience the combination or contrast of sweat and serenity. But isn't this awesome? I think it is. Now let's talk about parallelism. It's one of the very popular examples in marketing. So if you go to one of the websites that I've recommended to you previously, marketing examples.com slash inspiration, you will find a lot of awesome examples of parallelism. Let's have a look at what tragedy brought us here. Introducing our fitness studio where passion meets performance and sweat meets satisfaction. Can you see these parallel constructions within the sentence? Passion meets performance and sweat meets satisfaction. I think that's cool. I always state of the art equipment and expert trainers will help you achieve your fitness goals. Whether you're looking to build strength, improvement doings, or just sweat, join us today and discover the power of parallel progress. Again, it does make sense, but I don't want to use the word of the creative technique. So instead of parallel progress, I would say, discovered the power of harmonic. One more. Okay, asthma is that our fitness studio? You won't just sweat. You'll succeed with the state of the art equipment and top-notch trainers will help you achieve your goals and exceed your expectations. Whether you're looking to slim down or bulk up. We've got the tools and the know-how to make it happen. Come see for yourself and experience the difference that dedication and determination can make. Well, I kinda like it as well. You can see this like you won't just sweat, you will succeed. That's an awesome progression with. Now let's have a look at the sentence rhythm example. Write a paragraph about using Slack for fun in the following sentence for them. One short sentence to compound sentences. Short sentence. Slack is fun, brings people together, allowing them to share jokes, memes, and engage in a friendly banter. Colleagues and friends can create custom channels, use emojis, and even play games to break the monotony of work. Laughter ensues. Well, I would say unites. But anyway, it's just changing one word. So for me this rhythm really works well and it's readable, but still content for. Now, let's do the following thing that we've discussed previously in this lecture. Now, let's use a more complex stylistic device. In this case, it's an illusion. And when you have an illusion, you want to specify what the allusion is four, because this is like a reference to something that's popular or well known in literature, authors often create allusions to other authors, like ancient authors. In our case, we have the copy that we like. Now we ask charge a petite, we refine this one. The prompt sounds like this. Rewrite with an emphasis on stylistic device, allusion to metrics movie and I'm ending up with this one entered this dielectrics. Users dodge workplace, worried, I'm like Neil evades bullets, waving through a digital realm of humor and comradery. They choose the red pill of amusement, discovering custom channels, emojis, and games to escape the mundane unplugged joy. For me, that count as an allusion to Matrix movie. The slack tricks, the dodging workplace. Barton like Neil evades bullets for me that counseling. So let's summarize our lecture on using artistic and stylistic devices. With Judge EBT. We've learned how to use stylistic devices to add our copies, sound and structure. You've learned how to add contrast and comparison. And we've also learned how to use more complex contextual stylistic devices with Chad deputy. And as a bonus, we've learned awesome paragraph sentence rhythm. Use all of these tricks, learn about them and have fun experiment. So let's awaken our artists and bring some art into AI generated contact. 30. Turn One Good Idea Into Many Using SCAMPER Technique: Video, you learn how to transform one good idea into seven more and how each of those seven ideas can be extended into even more ideas. Well, based on the feedback that you've provided, generating content ideas is one of the most challenging aspects of your work, and that is totally understandable. Staying creative while consistently producing a large amount of content can be quite tough. In this video, I'll show you a simple prompt that combines a few prompting techniques and one powerful framework. Let's go have a look. The technique that makes this possible is called scamper. This is an acronym that stands for, substitute, combine, adapt, modify, put to another use, eliminate, and reverse. Each of these actions represents a way to take a source idea and generate new ones from it. The best approach is to start with some of your top performing articles, blogs, videos, or any content that you want to work with. For this demonstration, I'll head over to Tech crunch and pick a random title that I like. Let's choose this one. One to branch out beyond Google. Here are some search engines worth checking out. This will be our example. Next, we'll use CA GPT to start prompting, and let's enter this front. Act as a marketing content writer. Your task is to generate blog post ideas with headlines. The block is about generative AI for marketing specialists. Act step by step. Analyze the type and style of headline and why it works. Then apply the scamper technique and similar headline logic to generate new ideas. Here's an example. Want to branch out beyond Google. Here are some search engines worth checking out. The step by step instruction is crucial because it ensures CGPT thoroughly analyzes the original headlines. Okay. Now, let's look at the results. My opinion, the output is quite solid, The answers are structured with headlines that clearly follow the pattern, but cover different yet somewhat related topics. If you have specific information, you can tailor your topic even further. If you need more ideas, you can either pick a different example or refine one of the suggestions that you already have. Another successful tactic, in my experience is cross referencing content from different channels. For instance, if a hook works well on TikTok and attracts the right target audience, there is a good chance a similar approach will succeed on link ten. Now, if you notice a lot of repetitive cliche or that typical GPT language in the generated content, feel free to switch to custom GPT model tailored to your tone of voice. You might already have your own, or you can use a ton of voice guide like mine to refine the outputs. Can help you maintain the consistency and ensure that the content feels authentic and aligned with your brand's style. Okay, that wraps up this quick tutorial. Although as brief, this single prompt is a highly efficient way to generate more content ideas from one that has already validated and proven itself. 31. Creative Ad Idea Generation Prompt + Reverse Engineering: Have you ever struggled to come up with creative ideas for ads? Well, today, I'll share a structured approach I've developed that consistently generates powerful concepts for both print and video ads. This approach is based on reverse engineering technique. So I reverse engineered classic ads that I chose to my taste and then refined the prompt through iteration and experimentation. I'll show you how such prompts are created, and then we'll go to the prompt breakdown. And then we'll look at some quick examples for popular brands. So let's go to my screen and have a look. So now for this lecture, I created a little project within hA JBT so that I can show you a few examples and how we created this prompt. So let's first look at ad creation instructions. So first of all, what I did, I wrote a prompt that would work as a creative ad agency. I mentioned what I'm going to provide, what I want in the output. Basically, this is a reverse engineering prompt. Right here. You can see the process. You can see what is expected in the output and what kind of features, what kind of tools we expect the model to use. I also want to mention that I'm using GPT plus subscription, which is a paid subscription, and with the recent update, the most powerful Open AI model to date, which is a one is included here, and I used this model for this task. I also mentioned that I will provide multiple examples, so I can only upload ten files at a time. That's why I mentioned that I will provide the files in three portions. You can see a list of ads here, but you can actually provide any kind of examples that you like because in the prompt itself, we are not going to use these references and these examples. So you can see we have three portions of the prompt. Then we see a detailed analysis made by this one. You see the breakdown, and then you can see the prompt blueprint, the methodology, and so on. What we're interested in is the master prompt template in step four. And now we have the prompt here. It came up with a very interesting prompting approach, which I like to call semi agentic. So first of all, you always have this row, then you have a step by step instruction, and you can see across all of this prompt you have the characteristics of the style that we're going for, which is minimalistic, witty, and brand appropriate. What's important is that we actually have steps to follow and they are specified. So first of all, you have the brand essence here, so we need to summarize it. Then we need suggested five potential one liners. After testing, I realized that we get much better results when we ask for ten potential one liners. Then we select a visual concept that pairs well with this option. Then compose final add. What makes this thing like semi agentic, is that we ask the model to pick the single best one liner from the options that it already provided. Of course, if it picks the wrong one, you'll be able to change it. I mean, you can say I preferred a different option. But I noticed that a lot of the time, once the LAM has gathered all of the details about the brand, the choice is quite good. Then you have a description of the visual layout. And you have a check for brand relevancy and you have the final output after all of these steps. Quick disclaimer, this prompt has changed a lot and I will show you the final version, and also the final version is available in the prompt book. Most of the time, it fits within one or two regenerations. Let me show you where you can find this prompt in your prompt book. In your prompt book that's in the resource section of this course, you have this idea generation prompt here. Once you get to the top of the page, scroll all the way down towards creative ad idea generation. Here is where you see this final prompt. This final prompt, what I do for your convenience, you start the prompt by pasting the brand or product. For example, Google Pixel. If you create for a brand that's well known and if there's no training data, you can add the data within your project or custom GPT in a knowledge based document. In that case, the prompt will do the rest and it will fill out all of the variables. If you want more manual control here, or you work on your brand and you know the details for sure, you may want to actually fill out all of the variables yourself. Within steps to follow. We started with the same brand essence. I think that's a very good reminder for the process and for LLM. Then we asked instead of five, we asked for ten potential one liners. Then we extended the criteria for the one liners and as I experimented, I noticed certain patterns from HAGPT that I wanted to eliminate because first of all, they can be applied to any brand anytime. That is something that we usually don't want in marketing, right? So if you can say the same thing about every brand, then maybe this is not the right brand message. You want something more unique to be prioritized. So first of all, what we're doing is we're avoiding generic options smarter, not harder, one thing at a time, and like, something like perfection every time and anything with the word quality. Then we continue with the visual concept, and we add more acceptance criteria in the visual concept, and we emphasize that the visual has to interact with the copy and has to creatively amplify it. This way, we get more of a visual metaphor added to all of this composition. And then the finalization is practically the same. What I also added is an option to use Del E to visualize the storyboard. I don't like DL as a visual generator, as you know, from my past lectures here. But actually, once you're within the chat, you can create a quick reference for your designer for yourself to take an image or for a different image generator that you're going to use. So anyway, I think it's a great way to start and to get a reference for visual as well. Again, I don't recommend it as a production ready image. Alternatively, what you can also do to even amplify and make those results even more relevant try to attach a few samples, a few visuals that you actually like. Here in Notion, I added a Google Drive link to a little collection of the ads that I like so that you can just go and play around. I edit it for educational purposes here. But I encourage you to make a collection of your own so that the outputs that you get are more like you, like your preference. Alright, now I want to go to examples. So for some of you who don't know, I like to play guitars in my free time. In this industry, there hasn't been a lot of innovation for years. And now here we are with guitars that are amplified without amplification, so they have some sort of effects going on within the guitar, so you don't have to have an external amplification really cool thing to talk about. And these are titles that they have on their website. I thought that might be a good idea to go and create something for this brand as an example. Let's go to this example in our project. So it's the same prompt again here. Then here are the ten potential one liners that were created for this brand. I actually like more than one. I also like unplugged yet more connected, but I would use it for projects that are more community based. And I think that acoustic soul and electric brain is a very good reflection of what they dear, it may touch to a lot of musicians who are creative people and they are emotional, it refers to both rational and emotional here. We have this visual concept also here. We have a suggested sub headline. I actually like unplug your creativity because acoustic guitar players are keen on being unplugged and we have a visual storyboard for three shots. This could easily become a quick five to ten second video or just one image. I visualized it with the deli. Of course, the visual part is not at its best. But you can see the combination here. It could be adjusted to their images on their exact product. At my daily job, one of the projects that we're doing is actually generating images with the particular product database of the client so that the product itself is adapted to a generated image, but it's not modified by diffusion models, and the text doesn't go away. The product actually keeps looking like a particular product uploaded. Again, can this be a reference for a designer? I think absolutely. I've seen posters and banners and ads that were much worse than this. Let's move to one more example. Let's move to Google Pixel. Although I do use an iPhone, I think Google Pixel is a really great product. Same prompt here, and here are the generated one liners. So first of all, yes, it's true that Google Pixels emphasize a lot on the AI capabilities and the minimalistic design. I think it's quite true about these phones. A lot of focus on the camera and AI features. I think one of the features that I really liked when I played around with Google Pixel was an option to fix a photo that's blurred or not sharp enough. But okay, let's move on to the one liners. Of course, it played around with the naming, which is like pixel perfect. Nice hyperbole here, but I don't think it would choose it. Let's have a look. So pixel perfect every time. I'm not sure I like this. So let me go through one more iteration. I will just base the same prompt again. You could do this by regenerating the prompt. Now, let's have a look at other options here. You see the same option, but let's look at more ones. The other options seem a bit more legit this time. So for me, acceptance criteria is that I can get an acceptable idea within two or three generations. And here at the second generation, we have an idea that I actually like. I intentially don't hide it. I want you to see that sometimes you just have to regenerate to get a better result because you don't get exactly the same result every single time. Okay, now let's have a look at the final ad. So we have the logo placement. We have what's in the middle of the screen. It could be either a static shot with all of the elements at the same time, or this could be an animation. Of course, I would like it to be all clean. Terms of the fonts and branding and the photos of the product itself to be the exact model that I'm showing. But again, as a reference, it's more than good enough. What I would also add is I would make before and after effect within this frame. So I would make one part of this image more blurry. Let's look at one more example here. This example, I uploaded a couple of visuals, a couple of banners that I like. Here are they? And I just changed nothing except the brand. So let's scroll all the way down, and we see ten potential one liners here, and I like this one, but I thought maybe we should change up something choice that the model made and my choice weren't exactly the same. So in the next prompt, I said, too many words, I like UHD. I realized that this is not the best looking option. And when you look at the banner itself, you don't really understand the whole idea. So I played around with it again and again I decided that actually the initially suggested option was better. Ultra HD was actually a better option in my taste. Again, this is very subjective at this stage because I'm not working or I'm not affiliated with any of these companies. Let's go to one more concept here. Cannon. Again, the same situation, the same prompt, one liners, the choice, focus on what matters. Cannon is well known for their auto focus system, which is very good and actually a competitive advantage. Fok is on what matters. Actually, it is not a bad idea. Let's consider it not bad. Let's visualize it. I think that this visual part doesn't really amplify and play around with the headline. I really want more storytelling here and I just add it. The visual has to show an anecdotal situation. It's simple intuitive sentence. We already have a lot of prompting done before, so we can just refine with a single sentence. I think this tells a much better story. Okay, guys, I want to show you one last example. And in this case, we optimized the final visual part for a prompt for a visual generation. So we have the same prompt here. We have all the same story, except in the end when we talk about the visual concept, we have this specification and visual storyboard. We mentioned that we want shot by shot visual storyboard in the format of a mid journey prompt, and we want it to go from most important to least important elements separated by comma and have no more than 80 words. So we have the ten potential one liners here, and we have break free break better. Maybe, but I like SnapsmileRpeat more. We don't expect the choice to be always correct, but to my taste, this option is better. Okay, so I mentioned that I like different option, and we have the answer regenerated again. So we have this headline Snap smile repeat. We have a visual storyboard. So this is basically a prompt for a visual generation model. Please keep in mind that the Dobbie Firefly wouldn't accept brand name, so we'll have to remove it or use a different model. Okay, this is the visual that we get. Again, I think I've seen billboards, banners and social media ads that were much worse than this. And this visual, I would probably remove the smiley face and maybe adjust and regenerate a couple of times more just to get a more smiley face from the bite. Is pretty much it for this video. So let's summarize. So as you can see, this process can be easily adapted for image and video generation. For images, you can convert the visual storyboard part into a prompt format for Dali, Mid journey or any other visual generator. Dali is by far, one of the weakest image generators, so feel free to choose the one that you like. For a video in the prompt, you'll have to specify that it's going to be a video made from different shots and specify the amount of them and make sure that you give the format for a prompt and ask for a description shot by shot. If you want to go fully automated, you can create reusable templates and put them into project or custom GPT. And then in your prompts, you will only adjust the goal, the message, and the product of your campaign. See you in the next lecture. 32. Remove These Text Elements to Avoid Sounding Like GPT: Today's one-minute charge if the tip is about proofreading and editing your copy. For example, for social media post tragedy was taught that it's important to grab attention in the beginning of the post. And usually charged produce starts writing social media posts with something like attention and audience or exciting news, something, remove that. Now that nonsense, I mean, it's too generic and then it's obvious that it's Chad GBD, second tip. Watch out for sentence constructions with from something to something, as it's very typical for a child, jeopardy on my construction is not only something, but also something else. And whether you want to do something or you want to do something else. And yet, one more thing while editing those posts and sending them to stakeholders based special attention to the hashtags that show Jeopardy suggests and be really cautious about those emojis. Maybe try using emojis for navigation purpose. Not just spread them, whatever chad deputy suggests. Be extra cautious as this might give out a tax generated by AI might look not very trustworthy. Yeah, that's it. I guess I'm a bit over a minute, but I hope these were quite useful for the time that you spent with those steps. Cheers, Bye 33. L26 Train your own model within a GPT convrsation: Alright, welcome back everybody. In this lecture, we are putting all of the job that we've done before into one awesome, I would even say mind-blowing workflow. The workflow is called, Is that understood? The true beauty behind this workflow is that we are literally training our own API modal without any coding or technical knowledge. Just a couple of cool fronting phrases. And one cool approach, excited, Let's have a look. So what do we want here is to find a convenient way to prompt charge apathy, to use all of the information, all of the strategic materials, documents, researchers analysis, summaries that we have prepared to produce the relevant results. And GLUT4 actually is very good at it. The 3.5 modal is also capable of doing this, but I found that GLUT4 follows my instructions more accurately. Alright, let's have a look at the prompting logic. So the initial prompt follows our initial logic. We give a role than we describe the task. Then we inform what kinds of information we are going to provide. Then we give a negative rule, which sounds like this. Don't write anything yet. And then the question is that understood? The awesomeness of this approach is that judges for D says, Okay, give me your information, train me as your AI model. And I will use this information to provide you with more relevant copy. So we send that prompt over. Then our second, third, fourth, prompts are also a bit different. We start saying what kind of information we are providing at the moment. E.g. here is my audience segmentation, semicolon. Then you provide your audience segmentation. Then in the end of this prompt, we are saying, what information we're providing next. Then we say, don't write anything yet. Is that understood? And we have this iteration going on and on until we provide all the information that we want to learn before writing for us. Now, once we have given all of this information into judge deputy, our last prompt here is also very important. In the last item we say, now write, complete this task. Then we say, remember about the types of information that we have provided before, and that's it. That's the workflow. So now let's go to my screen and have a look at how it works in my charge of t. Right now on the screen, what you can see is a spreadsheet with the information from our strategy. I simply copied it from that place to store in one spreadsheet because it's easier for me to navigate like this. You will see this spreadsheet and the resources of this lecture. My personal preference for this type of content is to keep it all in spreadsheet or Excel format. Because I like to have it all in one document, but you can find whatever works better for you. For me, that's the option. And this option forces me to do one extra step once I'm here. So let's have a look. I'm copying all of this and I'm going into chat GPT. I named this chart formatting and I'm just asking to rewrite this in a bullet point format. I'm basing this information here in this prompt. Then I have the PESTEL analysis in a bullet point format. I'm doing this with all of the other strategic outlets that we have. From our strategy section. You can see I have all of them here. Great, and I'm finishing here with the content mix. Now let's go to our, Is that understood content plan. I'm going to show you how a content plan can be created with this approach, a named this chat like this, because I might want to revisit it again in the future. Or I might want to try a few more chats. And I don't want to repeat this procedure again and again and again. So let's have a look. Your content marketing expert, right, a content plan for two weeks. I'll provide you with marketing landscape, swat toes, strategy, audience segments, social media channels and rubrics. Is that understood? Then we have the conformation. And here I'm pasting my pastel analysis. Then I say, next, I'll provide you with swot toes. Don't write anything yet. Is that understood? I have the confirmation and then I say, I'll provide you information on my audience. Don't write anything yet. Is that understood? Yes, it's understood. Great. I'm pasting the segmentation. Then n form that I'm going to paste my content makes. Don't write anything yet. Is that our understood? Awesome. And I have this content makes here. Then I will provide with tone of voice. I mentioned the tone of voice. My tone of voice is casual and informative. According to Nielsen Norman Group. Than now use all of the information above to create content plan for each social media channel. Don't forget to indicate the target audience segment, channel and Rubric. Let's have a look at what we have in the output of this two-week content plan. So week one, Monday, Facebook, family **** is drivers. Rubric is for its history and milestones. Let's double-check if that's correct. Yes, I have this in my content links so we can move on. The content says, share your post highlighting for its history and commitment to safety, featuring an iconic family friendly vehicle from the past. Alright, no copy yet. No problem. No problem, right? Alright, to use the Instagram, luxury car enthusiasts and beautiful car photography and design. Both tonight getting photo of a luxurious Ford vehicle showcasing it's elegant design and high-end features, awesome TikTok, commuter convenience seekers, rubric, driving tips and tricks. Sheriff, a short video with tips and how to efficiently navigate through heavy traffic. Who are rights? That sounds okay. Feel free to pause the course and have a look. Okay, now, the topics seem legit. Let's move on. And I'm going to ask to LGBT to write texts for all of these posts. Asking once again, not to forget about the tone of voice and audience segments. Now, let's have a look at what we've got. So it's Monday channel Facebook, family Park is drivers for its history and milestones. So far, everything seems clear. Travel back in time with us for it has been keeping families safe on the road for decades. Check out this classical family friendly for a vehicle from the past. I mean, we could rephrase it and mentioned the modal that we know. Do remember taking road trips and one of these that sounds quite emotional. And then we have this, share your memories in the comments. Like, let's have a look at few more. I want to see first week TikTok, audience segment, commuter, convenience seekers, rubric, driving tips and tricks. Let's read this. Content, stuck in traffic. No worries, we've got you covered. Here are three quick tips to help you navigate through heavy traffic with ease. Try them out on your next commute and let us know how it goes. If I want more details here and I want this video script, I can use the prompts that I've provided you with an hour video scenario lecture, the video scripting one. I can use this. Now if I want to create a more detailed video script for this idea, I can go to the prompting list attached to this course and script this video for TikTok. But so far, it sounds correct to me. Let's have a look at one more. Let's have a look at Instagram and different audience segments, which is adventure seekers, rubric, adventure road trips. The content is ready for an adrenaline pumping adventure. Ford modo name is built to tackle the toughest terrains. So you can explore the great outdoors without limits. Swipe left for more stunning shots of this fraud beast in action. I like it. It's absolutely work, workable, not genius, but still something that you can regularly keep up with. Now you can also see here that charge a PDF finished writing abruptly here. So I'm just politely asking it to continue writing and we have it all done. So let's quickly summarize. This, is that understood workflow is here to help us train GPT-3 for in particular, so that we can actually get something way more strategically relevant to our gulfs. The difference of this workflow is that we asked jejudoga t, Is that understood? So it doesn't write anything right away and has some time to learn. And then we finally ask it to write something. We reminded of what kind of information to poor. I encourage you to experiment with different types of tasks with this workflow, e.g. if you want to write an SEO article, you may want to provide with your keywords that you've researched with Google Keyword Planner or a tariffs or SEM rush or SDI ranking or whatever, maybe bruce suggest, I don't know, may want to provide your keywords, the information about your brand, the information about your audience and topics you want to cover. Then you want to create an outline. And then you want to ask Chad GPT-2, right? Part by part, you can also create a landing page copy this way, many, many other things. I hope that you will be able to use this prompting approach as a glue to connect all of the work that we've done here and that we will learn to do in the future as well. Hope you're as excited about it as I am. Thank you and see you soon. 34. AI User Research: Introduction to Advanced Data Analysis in ChatGPT: Of understanding and empathy is one of the most popular reasons why startups and small to medium businesses fail. Actually, while talking to your users and analyzing their responses, you are very likely to be surprised. Believe me. Especially when you compare their responses to opinions of your colleagues, managers, and, of course, your own ones. Even though we're here learning about AI, I can't emphasize enough on the importance of human communication and connection. The approach I've prepared for you here brings this communication to a new level. So user research with GPT can be tricky. On one hand, large language models can generate quite reasonable assumptions, sometimes better than we do. And that's great. But what if we can access the actual users or target audience? Well, that is much, much better. In this section, you will learn how to plan user interview questions, how to analyze and process them. And I think this is a perfect use case because it uses the advantages of AI in terms of processing data and generating new assumptions based on it. It actually follows the formula. Data plus assumption can lead to good decisions and hypothesis. Since recording these lectures, Chad DPT has made a few updates in the interface. You may see a few differences, but if you can still see this lecture, then the features required for this analysis are still available in one way or another. By the way, quicker update, code interpreter was renamed into advanced data analysis. It was removed as a standalone feature, but it's natively integrated in CA GPT. And it's worth noting that most of the approaches discussed in this tutorial require a GPT plus subscription. But if you don't have it at the moment, I still recommend you that you go through these lessons. You never know until you know that you need it. Right. See you in a few seconds. 35. What Is ChatGPT Advanced Data Analysis (ex-Code Interpreter)?: So what Is Code Interpreter and why it's important, how it works and how to enable it in ChatGPT. As we discussed previously, ChatGPT has lots of limitations. You can't upload or download anything. For example, if you wanted to show it the diagram we're analyze something. It can make charts, it can run code in the interface. Well, Code Interpreter is here to change all of that. Code Interpreter is a new feature in ChatGPT, particularly in the for modal, which runs Python code in Python working environment. Basically, it's a Python environment where you can execute Python code to perform almost any task you, unfortunately, it's only available with ChatGPT plus and for me as a marketer, this feature finally makes much bigger bang. For $20 a month. It may sound like a feature built for coders, but it can also help us marketers with tons of Tasks. And surely we haven't uncovered all of the potential yet. But let's start. Well, the crucial ability here is to perform various tasks using your files. Yes, it can now upload your file and execute regular prompts with it. Finally, by the way, all prompts within this course keep working. Except that with Code Interpreter, you'll want to ask ChatGPT to refer to your file as an input. Additionally, you can ask for an output and ChatGPT will create a file or infographic for you. So yeah, it opens up endless use cases for all users. And I'd say, especially for Marketing Specialists. So here are some of the use cases to consider. Process and visualize data from market reports, classified keywords by searching ten from your Google Search Console. Analyze your marketing surveys and create user Persona from them. Met your tone of voice on the whole new level. Got ChatGPT to finalize some format your Excel files and create derived metrics from your data. And of course, yeah, ChatGPT to seek insights from your data. All of this either needed workarounds or wasn't possible at all until Code Interpreter was released. Now, are there any limitations? Of course there are ChatGPT Code Interpreter can't access Internet yet. Again, freeze or make mistakes when given too much information. And it doesn't really understand all file formats yet. So let's go and enable it in your ChatGPT plus. First, yes, you need, that's GPT plus account. Go to Settings open Better features enabled Code Interpreter, start a new Chat, choose GPT4 modal with Code Interpreter. And that's it. Code Interpreter is probably the most important improvement to ChatGPT so far. In terms of expanding marketing capabilities. Do you make sure that you spend enough time with it? And I'm sure that it will pay off. Personally. I'm going to invest a lot of time and it, and recording new lectures with new use cases. So I'm excited to see in the knee lectures, particularly with this update on Code Interpreter. At first, there will be a separate section regarding Code Interpreter. And when it stops being that new, I will just put all the lectures accordingly across the sections and the structure of this course. Anyway. See in the next lecture 36. ChatGPT Advanced Data Analysis (ex-Code Interpreter) tips : Now let's talk about some of the most important tips for prompting code interpreter so that you can apply them to almost any use case known yet. So first, make sure that the data you provide is already need accurate and understandable. Second, don't provide too much data, although I'm 100% sure that the year or two from now, the capabilities will be much bigger and it won't be a problem at all. But if you're watching it in the summer of 2023, that's the case. Next, don't upload any confidential data until code interpreter creates a data controller setting. I'm sure it will happen in future, but we're still in the beta. So there's no such feature yet. Next, when applicable, use chain of thought prompting, I think that we discussed earlier in the course. I found it quite logical and working smooth. Most of the time. You will want to ask to process and summarize the data first or do most of work you need in one take with one file. Otherwise, IGBT will forget the path to your file and you might get incomplete results. That's not much, but let's summarize anyway. In this lecture, we shirt some essential tips for using the code interpreter. Here are the main points. Ensure the data you provide is neat and accurate for the best results. Avoid providing too much data at once. Did not upload confidential data until charge if it introduces Data Control setting, use the chain of thought prompting technique for a smoother experience aimed to process and summarize data in one goal to prevent the loss of the file path. Remember to practice and experiment with the code interpreter to uncover its full potential. It will take time, but again, it will pay off and it's actually much easier to do as Open AI has increased the cap of messages that you can use with Djibouti for 250 within 3 h. Maybe at the time of watching this lecture, There's no limit at all. Anyway, I thought it's worth noting it. Okay, see you in a few seconds and then next lecture 37. User Persona Worflow with Advanced Data Analysis (ex-Code Interpreter): In this lecture, we'll learn how to turn a survey into a user persona. Unlikely, purely hypothetical user persona that you can generate in ChatGPT before any actual contact with your audience. This method allows you to be closer to real life. So what will you need to use this approach? First, you need an Excel file that holds an anonymous survey from the users of your actual product. In this lecture, we'll look at a Google Forms expert. Second, you'll need a GPT plus account with Code Interpreter on. This is optional, but you can use your favorite buyer or user persona, template or prompt. So here's our Workflow. Upload an Excel file with your survey. Use a chain of Thought Prompt to get GPT to learn an unlicensed bull insights from your survey. Then add a context. Then ask ChatGPT to learn your survey and create a user persona based on it. Now let's switch to my screen. Have a look. You might recognize the arm that you filled out in the beginning of this course. Let's have a look at it for this example. Ai expert at my Google form as an Excel file. Now let's go to ChatGPT and Prompt it. So our Prompt goes like this. You are a Marketing Research Analyst for an online course using ChatGPT for Marketing. I'll upload an Excel file with the responses of my online course students. You'll pull insights from the question R3. Now, you can see that responds a few qualifying questions. Now let's say it's a good sign. I want all of that boat together and Interpreter it. You can see it goes step-by-step again. By the way, I've checked the survey and I can confirm that it got the math and numbers right. In the end of the response, it has replied with the suggested future improvements steps that actually make a lot of sense. Now I should say it's the conclusion that I'd make from this question as well. Now that we've got all of this here, we can ask ChatGPT to create a user persona based on this survey. Please note that usually there's more than one year, so persona, so you may want to ask for three to five profiles, but in this case, there's not much data and I don't want to make this lecture overloaded. So let's just asked for one, by the way, Makes sure that you're prompt book is updated in July 2023 or later. In this update, I added more prompts for user persona so you can play around with them. But so far, Let's have a look at the user persona that we have. It's marketer, Mike has five years of experience and the marketing field. His day-to-day tasks are often time-consuming and involve content writing and Idea Generation. Else's spends considerable time on Audience research, content planning, marketing planning, right? Occasionally he does competitor analysis as well. Now that's something that would do occasionally. Actually, I should said that it sounds a lot like me to be honest. He's particularly interested in content marketing and strategy. One of his biggest challenges is generating Content Ideas. Micro is not new to ChatGPT. Which majority of you are not new as well. And he can understand the interface and perform the basic prompts. But he's always eager to learn more about it and improve his skills. Sounds, all right, however, to imagine mic even better, I'd like to hear a quote from him. And as I cannot call in and talk to him, I will refund the Prompt. Now asked to add mics. Here's Mix. Now, as we're in ChatGPT, let's ask to export this user persona in an actual file. You can see it did require one refinement on my site. But in the end of the day, it was still way faster and definitely more FUN than the usual copy and paste process like Betty, go through back-and-forth. Additionally, you can ask Mike some questions. I'd recommend to ask Open-ended questions, not asked for solutions or opinions, but rather focus on escalate experience. Alright, now let's quickly recap this lecture. You can use your quantitative survey that you've gained online to upload it into Code Interpreter and then analyze the insights from the data and ask it to create user persona. Typically, you would want more than one, but for the sake of this lecture, we just used one. Check out the prompts on building User Persona just for inspiration and you can use them instead of the refinement that we've looked in this lecture. And also feel free to follow up with your questions that you're interested in, that you were thinking out. If you are experienced and product management, you definitely know how to conduct a user interview. So be sure to ask good questions. Please note, I'm not saying that this person is surely real and one single persona, but it definitely gets you started and it's definitely better than a hypothetical user persona that one it or not, sometimes used and can also help you uncover the real person and prepare for real interviews with your real users. I hope you liked this use case. I was thinking of how to work with user persona for quite awhile to prepare it for this course. And now we have Code Interpreter, and I'm super excited about having it help us with User Persona, hope it helps you. And if you have any more questions, feel free to reach out into messages off the course. And I will be happy to answer your questions. And also don't hesitate to use the Q&A section if you feel like doing so. Alright, see you in the next lecture. Bye 38. AI User Research and Advanced Data Analysis in ChatGPT: Summary : A. Now you know how to conduct user research or even prepare for a sales call using CA GPT. I hope that you've learned something new or even structured your existing knowledge, or you've got some fresh ideas to experiment with. You're now ready to prepare for user interview, analyze it, and take action. You can even generate user personas based on the insights and input. As always, all of the resources, including my custom GPT and prompts are in the resource section. I wish you all the empathy needed to understand what your buyers and users need, and the desire to build products that you are proud of and people are happy with. And of course, all of this should result in profit. And that's my wish for all of you. 39. L27 Create your Custom Instructions: Hi everyone. I'm excited to finally talk with you about Chajipeti custom instructions. Let's talk about their benefits, their limitation. We'll have a few presets for custom instructions. The general Use 1.1 that you can customize for yourself, particularly enhance and automate your Gibt functionality. To make your marketing efforts even more efficient, let's start. What are Cha Jbt custom instructions? Custom instructions are a new feature still in beta added to Chagpti by open AI. And it makes ChagpTi consider your personal instructions for every prompt, both information about yourself and the way you want ChagpT to respond. The great thing about customer instructions for Chagapt is that they are available for all ChagPT plans. Both free one and the plus one can be accessed through settings and beta window on PT website. At the moment, it's actually the competitive advantage of Ch GPT as Google Bart and Cloud don't have this functionality yet. What are the benefits of GPT? Custom instructions. You can tailor ChagPT responses to your recurring needs. You can save time by typing less things all over again, which is already a bit annoying. Sometimes, it can improve your communication, make it more efficient. If you test really well, it can even make ChagPT sound a bit more like you. What are the limitations of ChagPT custom instructions? The main limitation that I can't handle with some of a workaround is that GPT custom instructions don't follow your instructions partially or completely. Well, the future is still in beta and we can expect it to work better in the future, but it's still beta can work better. All right, anyway, let's move on. There is no preset switch yet. If I have a couple of sets of custom instructions, I can switch back and forth between them quickly, and I can change them within a conversation. Sometimes that's a bit uncomfortable, but maybe in the future iterations that would be fixed. Open Y. Okay, let's move on. Your context may change more frequently than custom instructions can handle. For example, if you change the context within a conversation or you need your conversation for a few different purposes and a few different contexts, then I would turn the customer instructions off and then set the dynamics myself. Now we can define two different types of custom instructions and marketing that could be either very general marketing use so you can consolidate your personal level information. Organization information, the industry you're working on can help you maintain that consistency across all of the tasks that you complete. We'll have a look at how it works in a few seconds. And specific use you can customize for your very specific job function that's repeating. For example, for your PPC workflow or your copywriting or management, or data analysis. Here are a few tips for creating effective custom instructions. Maintain a singular focus when you take a photograph. In most cases, you want one object to be in the center of attention. So it's here, avoid merging multiple specific uses in a general purpose custom instruction. Avoid any contradictions and oximerance. Don't ask VT to be both analytical and perdubercreative at the same time because the responses will be very mixed and confused. I would also recommend to outsource your repetitive tasks. Think of things that you repetitively type into Chagipt or things that you currently do yourself. How can you allocate them into the customer instructions? You don't have to explain things all over again. The shorter customer instructions, the better. I suggest that you think. As an SEO specialist, think in terms of keywords. Optimize the keywords for better word to meaning ratio. You need to have less copy but more actionable words. More things that Chagpt's language model can stick to when choosing how to answer. That way, you'll have the most stable results. Actually, it's something that I noticed from many thought leaders in this AI area. When they share their custom instructions, they give long sentences, very emotional. But in fact, you could cut that sentence into three to four words that Chip would understand. For example, asking it to give no disclaimers. Yeah, don't forget the limitations and the amount of symbols at the moment. It's 1,500 symbols in each of the boxes are given here for a reason. And one more important tip, and I think that's the most important one. Test and calibrate every change that you make to your favorite custom instruction. Don't get tempted to change multiple things at the same time because that way you won't know what worked for you better as the feature still doesn't work 100% stably. Test your changes with ten to 15 prompts that you regularly do and refine them accordingly if you see that something doesn't work. Here's the structure of general purpose marketing. Custom instructions that we'll work on in this course. We actually have an example of such instructions can simply copy and paste into your Chagupet. You can access it in the resources of the course. I will show you exactly where. Here's the structure of a custom instruction that works for me. I tested a lot of them. In the first section, you tell about your experience, your background, your industry, and typical tasks that you will give. You can also say what kind of answers you prefer all of the time. But that would be something that I would ask you to customize for yourself. But later on in the second section, when you explain how you want apt to respond to you, you start by introducing the role, then you describe the formatting and response style, then describe the reasoning process that you want Gp to go through. That's actually the place for your favorite prompt engineering techniques that you can also actually get from this course. And there's a whole dedicated section to prompt engineering techniques in this course. Feel free to add those techniques into this part of customer instructions, especially if there's one that particularly works for you most of the time. For me that's chain of thought. Then you can explain what you want in each case. For example, you can do if conditions, if this, then do this. But don't go too deep with it because that might introduce a contradiction or conflict. When you ask for things that contradict each other, then last but not least, limitations. For me that's the most important thing because my flow with age Pet consists of more removing stuff and clarifying stuff then adding things. I want chat to refrain from its cliches. In the copy that it often uses, there are words that you knew that are sometimes used. But then with the release of the popularity of Chagpet, you see those words and phrases in every single post and it's such a marker of AI generated content. In the end, we'll discuss a couple of useful add ons that you can add on top of your customer instructions. But I actually removed it from the preset that I use, and there is a reason for that as well because I don't want to waste the tokens, but I sometimes go back and paste those depending on what I'm working on. Now, let's talk about making custom instructions. Well, custom specific use for yourself. In the first section, you want to describe your current situation, your profession, your product industry goals, background target audience, your activity by activity, your goals that you're working on, the typical tasks and maybe even KPI's if they can be understood by PT, Maybe without numbers because you don't want to ruin the confidentiality as an option. You can add what types of answers you like. The second section would go the same pattern, but you just want to customize it all to yourself. The role, the response style, the reasoning, the conditions, the limitations, the structure stays the same, the details would differ. I recommend to focus well more on your needs in this case and something more specific. For example, create custom instructions for PPC or for SEO work that you're doing. So keep in mind this structure, but just tailor it to more specific parts of the job that you're doing. All right, in the next video, we'll have a look at a demo of how our general purpose customer instructions work. I'll share my prompt actually taken as a preset from this course, we'll go to PT, Use GPT four, and have a look at how this works with customer instructions and without customer instructions. 40. L28 Custom instructions Demo: Here we are in the prompt book of this course, the resource that you can always refer to. Go back here, Let's go to the custom instructions for marketing specialists. Here are the custom instructions that I have here. Important part, we can describe the tasks that we typically give to describe the answers that we want. Our ability to understand information at a certain level. I don't want GPT to be forced to provide all the information in a beginner friendly form. I also describe what I'm going to be doing now. The most fun begins in how you would like P to respond. I give a role, then positive rules, the formatting. I want AP format, for example, for better visual and psychological perception of my content. I want the responses to be organized and marked up visually, more for my perception, because I understand whether the answer is relevant or not, or good or bad for me, and I can skim it really fast. I want the break points between the paragraphs and so on and so on. You can pass the course and read it, or you can actually go and copy and paste, and play around with these custom instructions, which I really encourage you to do. I added the chain of thought, prompting. I think it's very adaptable to many use cases that we can have When it's not possible and not needed. Pt won't trigger this custom instruction, the negative rules, I also use them because these words have become such a cliche of chip. I instantly see that it's a chat post. Uses these words in the post simply because, well, there's nothing wrong with this vocabulary, but it's just over used by ChagPT recently. Every single answer that I asked for, it just puts one of these words maybe as a signature or maybe because it's just so popular. I don't know, But it just happens and I don't want to sound like everyone else who generates texts with GPT. Let's move on now. Let's zoom in and have a look at the prompt. This is actually a prompt from the copyrighting section of the course, but I included the placement, the topic, the audience, and a couple of arguments to the prompt. You can also find the prompt in the prompt book. In the copywriting section right here. There are actually a lot of prompts. Here's the result that we received. Feel free to pass the course, to read through the response. I can already see some of the things that are stopper for me and a signature that the copy hasn't been even edited after being generated by ChagPt. First, the log everywhere, the rocket log, which is such a cliche in 2023 and we're heading into 2024 already, right? Leave your, your leadership potential. That's the cliche that I don't want here. All right. Understand the difference. Again, unlock, right, unleash. These are things that sound a lot like GPT to me. Let's have a look at the actionable advice that Chip provided us with. I don't see that really actionable, because delegate effectively, okay, delegate might be, but empower your team and maximize productivity is definitely not actionable. It's very abstract. Develop strong communication. Again, very abstract. Sounds good in paper, but in real world it's easier said than done. I'm trying to save this post a little bit. With my follow up, I want the desire section to be more detailed by adding a recommended read and to rewrite the copy so that the reader can instantly apply these techniques. A lot of words here, but let's have a look. Delegate effectively, empower your team to take ownership of projects. It's still not very actionable. It's not something I will do tomorrow when I come to work. Free up your time for strategic initiatives. Again, it sounds very motivational, but not as contentful and actionable as I wanted it to be. One more word from the list of words that are used in every age pet response. Ensure your vision objectives resonate with every team member. All right. Learning adaptability. At least we have one recommended read here, which is already something. It's not a bad read. Okay, now I want a short post caption to get the audience into swiping the carousel. It's not short, it duplicates what we have in the beginning of the carousel. Here, it's a pure duplication, word for word. I wouldn't do it myself because it definitely doesn't catch attention. We have one call to action, second cult to action, that's the same. We have the words from the stop list again again. All right, this is great that we have such a vivid example here, but for some reason happens to me all the time. I really don't want the same words to happen in every single post. I'm not saying this response is bad. This is actually something that you can improve later by adding your inputs. But let's have a look at the base. Can we make a better base so that we avoid fixing the same things? Let's go and have a look at the example with custom instructions straightway. I can see that the sentences are sharper. Transition from a, B, that's a classical move. Do last, not more. I can see that these sentences are catchier. They're shorter as headlines. They are better. I don't see that cliche and those words from my stop list everywhere, I'm sure that somewhere they will pop up interest. Okay. Know the difference. Marketing distance and managing people are not the same. That it put the caps to keep the attention right. The meat of the post. Let's have a look at the meat of the post. Three pro techniques for future leaders. Actionable insights to hone your management skills. Delegate, don't dictate. It's basically the same thing. It's great because it's comparable. The sign tasks that align with the team members strengths. Well, this is already much more actionable. I can see that it can be done tomorrow. Okay, there's a metaphor and I can see why. Because we have this in our custom instruction. Let's have a look here in our custom instruction that's this part triggered. All right. Emotional intelligence. Recognize and manage emotions in yourself and others. All right. Again, his said been done, but it's very specific because managing and recognizing emotions in yourself and others is already an actionable pattern that you can take. Number three says, regular feedback loops create an environment where feedback is welcomed and acted upon the action. There's the word from the list. I would definitely edit it, but it's just one word from my whole list. In the previous and in the previous option, there were a lot of them. It's less editing for me now, But I think we can still bit of a better answer here. I'm going to do the same follow up here. Make desire section more detailed by adding a recommended read and rewrite so that the reader can instantly apply these techniques. Master these three protechniques today. Immediate steps to elevate your marketing skills. No weight required. I would remove that. I like the M from APA formatting. Love it. I can see that it's followed. Quick tip. Identify your team's unique skills today. Assign tasks accordingly. Tomorrow, it already uses this rhetorical pattern with today and tomorrow we have the recommended read here and it's relevant and there's this instant application. That's what I was missing in the previous response without custom instructions. During your next team meeting, open a discussion on individual strength. Assign one small project or task aligned with the strength that is really valuable. And I think that could have actually make it to one of my posts in the future. Yeah, there's this analogy from my custom instructions that I didn't ask for right here. But you can see it from custom instructions. I think it's quite usable. It's easy to memorize such content actually. Now, emotional intelligence. Quick tip started Journal already very actionable. Great read. I actually recommend to follow Travis Bradbury on Linked in Writes great stuff. Take note of situations today that trigger negative emotions discussed with a mentor or coach to identify proactive coping strategies. Great, love it and feedback loops. Quick tip introduced by weekly feedback Friday. Thanks for the feedback by Douglas Stone and Sheila Heen. Schedule recurring bi, weekly meeting dedicated to open feedback. It's something you can instantly do when you come to work. Love, it makes such a better post. Now let's write a short caption to get the audience into swiping the carousel. And you can see that this one is actually short. There's a hook that's different from the one that's on the carousel. There's one call to action without repetition, and I don't see anything from my stop list. Love it guys. How do you find these custom instructions? I think this does a much better job with custom instructions, especially when you use GPT and GPT for model. These custom instructions work seamlessly with the prompts that we have in our prompt book. I'm excited to hear your feedback on how that worked for you see in the couple of minutes in the next lectures. 41. Custom Instructions Customization, Addons and summary: The section two would have the same pattern, the same structure. But again, be more specific about that use case. 42. Create Custom GPTs: Instructions: This video, we talk about custom deputies instructions. As you begin working with custom deputies. This is the most important part of building your own put. This video contains some of the official recommendations by Open AI, and also some of the information that I caned with experience of building my own custom deputies and using the community based custom deputies. By the end of this video, you will be able to write instructions that make your custom deputies perform ply, reliably and accurately. Without further ado, let's start enhancing your instructions for custom deputies. Start off your custom deputies instructions by assigning a role and a goal of the chat bot, as well as some of the basic contexts, but don't overload it too much for now. Next, divide the multi step instructions into simpler manageable steps. This helps the model follow these steps more accurately. However, in GPT instructions, you'll need to break down the steps a bit differently than you would in a normal prompt. Normally, you'd need to say something like when the user enters this, then first do this, then that. And this can become confusing and messy for the algorithm within the custom gibt context. So instead, use trigger and instruction pairs. Let's break down a quick example. Type trigger, user enter start, then instruction, ask following questions. Then trigger, user provides answers to the question. And instruction would be to analyze the context, search the knowledge base for the most appropriate framework for the user's situation. Next up, use shot prompting, clearly marked instruction sets and call outs for a few shot examples to just make sure nothing gets confused. Providing examples, you can get even more consistent responses if that's what your use case demands. Now, if you're using knowledge files, which I encourage you to do, provide explicit instructions on using them, include specifying the file names at which stage and in which case, they should be researched by GPT. The same thing applies to browsing. Browsing is needed. Mention at which stage with the phrase, use browsing to do something. So by following these guidelines, you can optimize the performance of your custom GPTs and ensuring they produce reliable and acts. Now, I'll turn on some music, and let's look at example of my copy editor GPT Instructions. I use it quite often in real life, and you can, as well, I'll share the link in the resources, or you can just type the name that you see on the screen in the GPT store. All right, so that's it for now. See in the next video. O 43. Create Custom GPTs: Knowledge: Hey, welcome. Today, we will explore the knowledge feature in PTs. The feature is designed to enhance the performance of these language models and make it so much more personalized. This feature allows us to upload files containing additional context, which PTs can then access when responding to our queries. So how does GPT knowledge work? L et's break down how this feature operates. So you can use GPT editor to attach up to 20 files. Each file can be up to 512 megabytes and contain about 2 million tokens. While you can upload files with images, only text is processed, so images are ignored. Don't bother. It's just a waste of tokens. Actually, from my experience, I would recommend to upload the least amount of files possible to avoid any confusion. If you absolutely need multiple files, you'll need to provide instructions on when to parse which file. Let's now talk about how Custom GPT process knowledge files. So the GPT takes the text and divides it into smaller parts. Then it converts these parts into a form that GPT algorithm can understand. These are like code chunks. Then these parts of text are stored for later reference so that GPT can access them more consistently. Then when a user interacts with your GPT, it can access these stored files to provide context to the user's query. However, They cannot download the file, but they can get information and request information within those files. So please consider this as privacy feature. Now, while processing your documents, GPT selects either semantic search or document review depending on the situation and Pront. I like semantic search more because I believe that it works a bit more accurately in terms of retrieving the accurate information and reducing the amount of hallucinations. But let's talk about both. So Semantic search retrieves relevant text parts. It's ideal for Q&A style prompts where you ask for specific information from the document. Pad for product information reference, like features, pricing and so on, corporate training data or just your courses or notes that you upload as guides for your future reference. Now, let's speak a bit about the document review way of retrieving data because I think that in the future and coming iterations of CGPT, it's going to be massively improved. But it's just my bet. Anyway, document review reviews the complete document. It's best for summarization, translation prompts, or requiring the entire context of the document. In this case, for now, it's best to keep a relatively short single TXT or Doc X document that's neatly organized with special attention to keywords and headlines, no fancy formatting or whatever like that. Now, for a minute, let's talk about when to use custom DPT knowledge base. Well, it's best to use knowledge base for context that don't change all that frequently, or you'll spend a lot of time managing them. But think in these directions, like employee handbooks, policy documents, marketing strategies, product information, tone of voice guidelines, or any guideline or documentation that has to be accessed and referenced, frequently, especially by multiple team members. For example, right, it's cool to have this tone of voice accessed by four of your team members who work on the same channels and style of copy. Great. Now, let's proceed with tips for maximizing GPT knowledge based accuracy. So Step number one is formatting files. The Parser works best with simple single context formats. So avoid any overloaded docs and PDFs. By the way, PDF. Yes, they can be uploaded and compatible, but avoid those multic PDFs and complex layouts. It's also true about PowerPoint slides, basically. Then principle number two would be to guide the GPT behavior in the instructions. You'll want to use the instruction section in the GPT editor to trigger the search through the uploaded files before generating it from the base model or searching the Internet. All right. That's pretty much it. Thank you so much for joining this session on GPT knowledge files. By understanding these principles of building custom DPTs, you can effectively use this feature to enhanced capabilities of GPDs. So I worry if it sounds a bit too much at once. We'll look into examples, and once you see it in practice, it's going to be much easier. And then once you create a GPT or two, you will exactly know what you have to write and do to create a new GPT in the future. Thank you so much for watching till the end and seeing the next video. 44. CustomGPTs Other Tips: So, guys, in this video, we are going to discuss a couple of best practices that didn't fit the previous lectures, but I still think they're really critical to creating great custom pt, and they make a lot of difference. Without further ado, let's start. Tip number one would be to avoid over reliance on the GPT Builder. The GPT Builder is probably useful to someone who has never used custom GPTs, and it has a lot of limitations. So it tends to overwrite and adapt to your instructions, making it unreliable for defining the exact behavior. So instead, use the configure tap to craft precise instructions, especially if you already know something about prompt engineering, and if you watched the previous videos as well. The next step is more of a mental slash emotional thing, and I see a lot of people struggling with it. So creating a functional custom GPT involves numerous iterations. So be prepared for cycles of testing, analyzing the results, refining the custom instructions, and knowledge base. Start off with a simple use case and then gradually build the complexity. Don't try to make something genius right away. This iterative process ensures that your custom GPT provides a huge value consistently over time. Okay, so if you actually ignore this step, I have next s tip for you. And this is do not overdo. So avoid trying to make one GPT handle too many tasks. Focus on one task and define clear acceptance criteria. If you need multiple use cases, consider creating a pipeline of deputies. This is possible by using multiple deputies within one conversation. So you just add a custom GPT to your site panel and tag it just as you would tag someone on social media. Let's have a look at an example. Want to use a custom copywriting GPT to create social media copy in your ton of voice in your topic. And then you might want to create a prompt for an image generator because Dali is not a good image generator. So once you have this copy, what you want to do is just tag the mid journey prompt writer and ask to create a prompt for your post. And it will read this context that you have right now, and it's done. Then you can go to your image generator and use this prompt to generate an image. One more example would come from user research. You can use a user research GPT to generate interview questions, then summarize the responses once you have them, and then analyse the responses using a different GPT to visualize them and get better analysis. For example, using wel Form Alpha GPT. What I like about this approach is that it requires planning your process and identifying how multiple PTs can stack together to solve your repetitive tasks and workflows, therefore, increasing your productivity. Well, you see, productivity is producing more volume and less time and effort. And maybe with a bit more fun. If your use case is common, search for existing community GPT. There are great GPTs built both by individuals and by companies. But of course, for specific needs where specific data or guideline or tone of voice is needed, then creating your own GPT makes a lot of sense. But for general tasks, using existing solutions can save a lot of time of effort at the sacrifice of some personalization and individuality. So the next one is also a bit mindset related. You need to somehow decide whether your custom PT is good or not. So to evaluate your custom GPT effectiveness, if you work with the team, perfect, ask your team to share conversations they had and provide feedback. Therefore, you will on one hand, see what kind of steps they were trying to make and maybe try to improve your instructions this way, or just provide just use their feedback to improve it further. But if you don't have a team, there are also a couple of metrics that you can have as guiding lines for improving your custom dputes. And in my opinion, these are the following. One, your prompts become shorter, then you have to use less prompts in general. And then your answers become more detailed, accurate, and relevant, then your conversation with the base model. That's pretty much it for now. Let's quickly summarize. So avoid relying on GPT Builder, use high quality, well formatted knowledge files, test, and iterate continuously. Define clear acceptance criteria and goal for your custom PT. For general needs, use existing solutions or at least try looking for them. And of course, combine different prompt engineering techniques. For example, you can combine role playing, chain of thought, shot prompting, all of these work perfect together. And speaking of prompts, think of this. When creating instructions for custom GPT, your prompts will have to match and complement the instructions of your custom GPT. So you may want to think through this logic, so it happens naturally. Right. That's pretty much it for this lecture. Thank you so much for staying till the end. And I look forward to seeing you innovate with your GPT, coming up with creative use cases. And if you want to share your custom pt, please send a link to your custom PT and paste your instructions. And I will be glad to have a look. Also, if you have a conversation with your custom put, and you want someone to provide feedback. I gladly will. Anyway. Thanks again and see in the next lecture. 45. Get Started with Powerful Generative AI Tools in Each Modality: I still only associate artificial intelligence with Cajupt, it's time to explore the world beyond that. With thousands of tools at your disposal, It would be a sin to use just one. How do you figure out and find the tool that will best handle your task. This video is full of spoilers, but it also gives an actionable overview of the most important tools out there. Let's start with the basics. The artificial intelligence we mostly talk about recently is generative AI. It generates new information based on the data used to train the models and information entered by the user. This information is also called prompt. AI has certain modalities. For example, text, images, audio, video, and code. And this tutorial we'll discuss the specifics of these modalities and consider the best tool and their applications. So let's start with the most basic one and most common one that is text generation. Text AI models such as Chagpt are trained on huge amounts of data to understand and generate content. Their value lies in their ability to understand the provided content. Make conclusions and create clear human like text. Naturally, text is also one of the simplest communication media, so it's a great starting point for prompts to create content in other modalities like audio, image and video. Therefore, you get tools in the following categories, text to speech, text to image, text to audio, and text to video models, and even text to code. However, these modalities have their limitations. They can hallucinate, which means they can present incorrect or false information and do so very very convincingly. Therefore, they need to be controlled and properly conditioned. At the moment of recording, according to the user benchmarks, the best text model is Cloud three pus Panthropy and it's consistently competing for the first place with GPT four. It's a paid motel, but in my opinion, even the free version of cloud is powerful enough and sometimes definitely not inferior in quality to the paid version of Cap. Compared to the free chap anthropic clot is definitely more advantages in terms of argumentation and the naturalness of language. It's text just sounds more simplistic and natural. At least to me. Additionally, it has good optical character recognition, OCR, meaning it can understand handwritten text, PDFs, and image format. Logos in photos and details in photos. So you can upload an image and ask to generate captions about something, and it will be super relevant. By the way, if you work in notion and use its AI text generation, You're already using Cloud. The second most popular and one of the best text models is the paid version of C GPT, which is GPT four. It's huge advantage is the ability to personalize through custom GPTs that will generate content according to your unique requirements. Alternatively, you can use ready made GPTs from GPT store. I'll share the GPTs I built myself. And also some of my favorite PTs from the GPT store in this tutorial. On top of that, the paid version can be enhanced through third party services and automation, such as ZPR or mak.com or relevancia. Another advantage of GPT four is the code interpreter. It's a powerful application for working with code and data analysis. For example, if you have a survey of 1,000 respondents, You can upload the file, and GPT will group the results, analyze them, and then you can even prompt it to make data driven assumptions. Compared to other alternatives, the free version of GPT, which is GPT 3.5 is inferior. As this version often ignores instructions has limited functionality and generates sometimes very superficial answers. All right. One more tool to consider in text JNAI is Google Gemini. Its advantage is the integration into Google services. With its help, you can easily search for information documents, navigate email, analyze text, video, and more. Moreover, it has a great feature. It checks for hallucinations by verifying the results using Google search, and it also provides several response options for your comparison. But talking about OCR, remember what that is. Yeah, optical character recognition. So Gemini does not work with photos of people. Unlike Cloud does. The paid version of Google Gemini is also tempting as it provides a smarter, huge token size, also known as context Window, and also additional 5 terabytes of cloud storage as a part of Google one subscription. I need a lot of storage, so I'm considering this one for permanent basis. And all of that comes at around $30 depending on your country. You can also go to this website over here and see the current leader board among text large language models. Here, you can also test models side by side and choose the one you like more. Please do not gas light this resource as your free GPT. Don't tell it to anyone, and don't abuse it, because this gives our community a chance to choose the right model without paying for all of them, running all of these models implies costs that may get quite significant. So let's be respectful of what this organization does. Okay. Let's move on and talk about image processing and generation. Currently, we're talking about generating images from text, images from images and images from sketches. Which is actually part of images from images. These tools can be divided into in painting and out painting functions. In painting functionality allows you to change a certain area of an image using tax commands. And or selection tools. Out painting generates content around the image. For example, if you need to turn a vertical poster into a horizontal one or vice versa, AI will add the appropriate background for a larger format. Also a popular use case is background replacement. It's available in photoshop now with a Dobbie firefly three. But there's also a tool called photo room that allows you to turn one product photo in a product photo set. Very handy for e commerce and retail. Additionally, you can transfer structure or style from a reference image to the target image. Unfortunately, all existing models currently struggle with text. It's actually one of the reasons why me and my team at R&D built an enterprise tool that aligns with brand book, target audience and generates image and text separately and then puts them together seamlessly so that the image and the subject in the image and the text don't overlap. So what are the tools for image generation? So, most of the tools have similar functionality, but differ in result quality and various licenses. One of the most popular models for generating images is D E from open AI. But honestly, There are many newer and more advanced systems, which I will talk about later. Di, in my opinion, has relatively lower quality at the moment. I often generates less realistic images with a very recognizable cartoon like style. But it's easy to access and use. It's available in Chagpt Ping AI, and Microsoft copilot. Currently, the leader among AI tools in this modality is mid journey. At least, from the point of view of image quality. It operates on a discord server, which is not very convenient from user perspective. But the company is already testing a convenient web interface and the ability to customize various parameters of generated images. And it's also accessible via API. Thanks to its powerful algorithms, mid journey provides high quality results, although it is quite expensive compared to some of the alternative tools. And let me tell you a bit more about my favorite tool, which is a Dobe firefly. It's currently available for free as a part of creative cloud license. It's also nicely integrated into Dobe products. Has an intuitive interface with a wide range of tools for editing images and generating them. It includes effects, composition, focal length, lighting, styles, referent images. It's just perfect. I love how Dobbie dose works. Thanks to such an interface, in my opinion, it's the best way to learn to generate images. On top of that, their models are trained on a Dobbie stop data, so you can generate images for commercial use. Once again, it's getting integrated across all adobe products like Photoshop Express, and premier. So if you're into all of this suit, you'll benefit from a Dobby fire flight the most. Next one is stable diffusion from stability AI. Its interface is not as convenient as firefly, but any company or developer can download the source code for free and adapt it for their needs. Also, there's an interface called Dream Studio and it's available in Leonardo AI. The fact that it's open source opens up wide opportunities for integrating image generation technology into various applications and services. Thanks to an active community, stable diffusion regularly receives new features, optimizations, and bug fixes. By the way, there is a custom GPT that integrates stable image generation model, and you can integrate it yourself if you want to as well. All right. Let's move on with video and animation. The main way to work with video remains text to video, but there's also video generation from video, or video from pictures. Runway and Pickle labs are currently the two leading solutions on the market, specializing in video generation using artificial intelligence. They have similar basic functionality, although runway stands out for its greater flexibility and capabilities. In runway, the user can not only enter text prompts for video generation, but also directly draw or upload images in the interface, defining what and how exactly should move in the video. This allows for more precise control of the movement of objects, characters, cameras, et cetera. Collapse, on the other hand, is more focused on ease of use and automation of video generation process based on text instructions without additional settings. But I assume that's going to change quite soon. Both tools also support video generation based on existing video content. The user uploads the original video and AI aplos visual effects, objects, or animation according to the text description in the prompt. I'd also keep an eye on firefly from adobe, doesn't yet have a wide range of video features, but it's promising to become capable of removing some objects from videos defects, create storyboards based on script and lots of other cool stuff. The development of AI tools is a priority for a Dobbie right now. I'm confident that their functionality will expand and explode, especially based on their collaboration with open AI, the coops, and runway. But, you know, video is a unique modality in terms of content generation as it contains music, text, voice, images, sub titles and descriptions. So lots, lots of stuff from other modalities. Therefore, many tools integrate video processing technologies instead of full fledged video content generation from text. In this perspective, one of my favorite tools is descript, which uses speech recognition, which is like speech to text. Model. To transcribe audio from video like this one, into text format, The transcript synchronizes with video track, allowing for easy video editing. Right in the transcript, you can cut out unnecessary pauses, repeats, or failed takes. Actually, just run one click. All edits are automatically reflected in the video without the need of tedious, manual cutting in hours of boring work. Similar functionality is also available in premiere, but descript is much simpler for those who are not professional video editors. Okay. The next modality is audio. So audio modality has two directions. Speech generation and music generation. Generating audio from speech to text, which means transcribing speech into text then speech to speech, converting speech into one language or another language or another voice or language. The descript app that I mentioned before, In addition to transcribing, also allows coning your voice and generating a voiceover through text to speech technology. Additionally, I use the tool called 11 Labs, which allows you to work nicely with your own or with a proposed voice from the community. You can type the text or read out loud and generate audio with another voice, choosing the tone, accent, mood, and more. This is very convenient when there is no opportunity to record a voice over, or there is just no good microphone round, or you just need to improve the sound that you have. For music generation, you can use sono, which creates songs with vocals based on text descriptions and instructions. But in practice, I use stable audio 2.0 at the moment. That's the latest version, and I use it for generating instrumental music like the one you can hear right now. Okay, so you have text to audio and audio to audio in staple audio to point out. It means that you can write a prompt or hum or sing a melody, or a beat to get music from it. It allows you to create a track up to 3 minutes long choosing the actual precise length, style, instruments, and moody need or added a track that you up. It's ideal for background music, for videos. It's especially convenient that you can manually determine the length of the track. It means that your audio track for your short up to 3 minutes long video will have the right dynamics and will not stop in the middle. By the way, stable audio also has a separate page with fronting guide. It's quite simple, even if you're not an audio file. But, of course, if you are, you'll benefit from it. Okay, that's pretty much it. Let's navigate over the few more tips around GNI tools to summarize. So there's no universal tool that would work equally in all modalities. It's best to choose different tools for different purposes. How do you do that? Here are a few more tips to help you choose the optimal tools. First, formulate the tasks that you want to assign to AI. Think, which repetitive work with other processes It can be automated or improved with AI, and then just approved by you. After that, determine the necessary modality that suits your task. It's better to choose a tool that is strongest and certain modality or incorporates multiple tools. To achieve one goal. Focus on personalization and automation. These are two main benefits of AI. Look for opportunities to customize the tool to your needs, style, and data, and of course, automate repetitive routine tasks and processes with I so that you can in the end just approve or edit it slightly. But while automation is good, please please please stay away from unsupervised use of NAI, especially for content creation. Just stay away from the idea of creating 500 posts in 1 minute. Just please Just trust me. You don't want that. Be critical of universal multi functional tools that promise to do it all for you. In most cases I've seen, these are just containers and interfaces for CGPit. However, there are solutions like the script that I mentioned before, and they integrate several different tools to save time for a specific use case. This one is counterintuitive, but invest in real life experience. Explore cases, communicate with colleagues with people who have more experience than you do or who have different experience. Attend various conferences, master classes. Travel. Learn to distinguish good from bad. I bet you know how to do it without this training, so it's not the focus of this course. And last but not least, you don't need hundreds of tools. Instead, learn the ones you like really in depth and look for ways to creatively chain them. You can even automate some of the tools using make.com, CPR or relevant CI. I regularly use round 20 applications. And in addition to those I mentioned previously, I use perplexity ai.com for information search. Sometimes I use GM app for creating presentations or Cursal posts, and there's also one of my favorites Harp AI. Which can be added as an extension to your browser to analyze pages, YouTube videos. For example, it's great for learning, content repurposing, summarizing information, You can analyze keywords from articles, especially when it comes to LSI keywords, sort or Gmail, interestingly, you can even monitor updates of a certain component on competitors pages. Say the price of a product, or you can do it for yourself, not just work. Remember this video on the eve of Black Friday. Anyway, AI is just an assistant. Do not lyly trust it and rely on generated results 100% without any supervision. On contrary, try to be more interesting than the default CPT response. Use AI consciously and always critically evaluate its work as at the end of the day, you are responsible for the final result, not. All right. See you in a few seconds. 46. Integrate ChatGPT Into Google Sheets or Excel: Welcome back, everyone. By the end of this video, you will learn how to add Cage PT into your spreadsheets. Adding CGP to your spreadsheets can help you augment the already vast opportunities of spreadsheets or Excel. There are multiple ways and apps to do this, but I will share the most reliable one that I've tested. First, let's talk about why you might want to integrate GPT into your spreadsheets. Well, first, it's extremely powerful at working with text. You can do translating, extracting entities, tagging, categorizing, correcting grammar and spelling, and cleaning up data and formatting. You can create taglines, headlines, add copies. Product descriptions, subject lines for e mails, and drafts for blog posts. Using gibt integration for sheets increases your productivity by allowing you to get many variations quickly, Save answers in sheets for easier retrieval, avoiding constant running like back and forth with copying and pasting and CGP, and benefit from all the Google Sheet features, including real time collaboration. Basically, this is where you get rid of the repetitive monotonous tasks. If you have the same prompt, but multiple variables, that would be your perfect use case for using CI GPT and spreadsheets. Let's discuss some of the use cases. Say you have a prompt for writing SEO titles and meta descriptions. You can paste your prompt and put variables in certain cells of your spreadsheets and scale it as much as you want down the column. Okay, Don't worry if that sounds unclear a bit, right now, we'll walk you through step by step in this video. Discuss a few more use cases because it's really important for learning. So categorization. If you have social media comments parsed into spreadsheet, you can use GPT to categorize them into say three categories, positive, negative, and neutral. Or if you have a restaurant and collect reviews, you can categorize those reviews into service, kitchen, and other items. Translation. If you need to translate certain pieces of text into multiple languages, you can automate this process, even up to 80 languages if needed. Data formatting and cleanup. Let's say you have names, e mails or links parsed into spreadsheet in random inconsistent formats. You can use this integration to keep the formatting consistent. You can also use GPT and spreadsheets to interact with images. This can be useful if you need to generate multiple product descriptions or SCO old image descriptions? Yes, you can also use images here. And it's awesome. Web browsing is also a feature available in this integration. So if you have many browsing requests and you need to parse information from many browsing requests with different variables, you can scale it in spreadsheets. So is it worth to learn about it? I think yes. Does it help your productivity? Absolutely. With this motivation in mind, let's go ahead with a step by step guide. Step one is to install your extension. You'll find the link at gptfow.com. We're in the resource section of this course. Next, go to your Google Spreadsheets and find it in the extensions list. Find GPT f work Extension, and click Enable GPT functions. You'll see a pop up on the right. You can create your API key or pay for API usage to use all of the most recent paid models by GPT. But in my experience, DPT Mini was more than enough for these spreadsheet tasks. So let's go ahead and choose DPT Mini model. Now, let me quickly show you what the extension has. Basically has GPT functions and bulk tools. In the GPT function tab, you'll see tips, best practices, and list of functions. Click on the list of functions to see a huge list. Click on any function, to see an example, documentation, and even a video guide on how to use it. These guides are like a couple of seconds up to a minute. They're very short and super useful. Great for referencing, whenever you need it. Why suggest whenever you need a certain function, just click on an example, copy it, and replace with your pront. Or if you need to watch that video guide. You'll also see settings here, the safe mode, auto replace formulas, model choice, and even custom instructions. Yes, you can use custom instructions to define how creative the answers should be by adjusting the temperature. There's also an assistant that will help you generate formulas based on your text description. But if you're confident in Excel or spreadsheet, you probably don't need that. However, you can also get a formula explained, which is pretty handy if you're working with someone else's spreadsheet or you work in a document that you haven't touched in a while. Let me show you how to use the GPT function so that you have the logic for most cases. Once you've enabled the GPT function, chosen the GPT model and edited any custom instructions. Let's do this simple example. We'll take a short list of cities and create a fun fact for a stand up comedian to open the show with. In Column A, I'll make a list of cities. Of course, I'll start with the city where I was born and raised and where I live at the moment and where I actually record it now, and then I'll add Paris, Vienna, Amsterdam, Madrid, Barcelona, and Milan. Column B is where the magic happens. Let's create a formula. We type equals PT, open brackets, quotation mark. I'll paste my prompt. Act as a stand up comedian. Create a punch line for stand up comped opening. Make sure the punch line emphasizes the most well known fact about the city. The punch line should be under seven words. Let's close the quotation mark. Coma, and then I'll pick A three, which is the first city and close the bracket. Great. Now we have that first one worked out. Then I'll press Option on Mac or Control on Windows, go to the corner of the sale and drag it all the way down to the last city. Now, we should wait a bit, and we'll see all of these rows with punch lines that follow my instructions. Of course, In a real world example, I'll probably need to add an example of a joke. In my prompt, but let's keep it simple for the purpose of this video. Now, let's go to Bulk tools and have a look. You have Translate, extract entities, classify and search. Let's click, Translate. Here's how to set up the translation. Pick, Translate, sell in column. Let's pick B. Detect language. Okay. Then to Spanish and put result in column C. Let's add some translation instructions, for example, act as a professional translator here and translate from culture to culture. You can also add any source words like glossaries and set up custom instructions. There's a lot of flexibility here. Unfortunately, I don't know Spanish yet. So let me know how good this is. I will double check it by using a reverse translation. So I'll do C to English to D. I'll choose all nine raws and run the action. Now, let's compare the translations. I think it's pretty accurate here. Of course, we can have some variability, but it's quite natural even when using service of a professional translator. And of course, you can translate to another language. This two offers extensive functionality across various platforms. It can be integrated with Google Docs, Microsoft Excel and Word, and even your G mail. Yes, I G mail, it can be embedded to help craft and manage e mail responses. However, for me, the most valuable use case is its integration with spreadsheets, where it significantly enhances productivity and data management. For more information, visit their website to explore additional functionality or to find specific features that you need. The tool offers instant guidance, making it quick and easy to use once your setup and familiar with it. Let's wind up with a quick recap. CGPT integration in spreadsheets enhances productivity by automating repetitive tasks and enabling text based operations. Key use cases include text processing, content creation, data categorization, translation, and formatting. The integration allows for scalable operations, handling multiple variables with a single prompt. The tool is versatile and compatible with multiple platforms like Google Doc, micros Excel and Word, and Gmail, but spreadsheets are still the most valuable application. Okay. Hope this was practical and productive lecture for you. Not too boring, I hope. And I really hope that it will increase your productivity. See you in the next lecture. 47. AI for Research: Perplexity AI Feature Overview: Research is integral to our daily lives. We rely on it to make informed decisions, solve problems, innovate, and learn. Traditional research methods can be time consuming and require extensive information, gathering and analysis. CGT browsing has also been a frustrating feature for a long time. They launched it, then it didn't work. Then they unpublished it, and then they published it again. Now, it works. It's not that bad, but it's also not as accurate and advanced as research specific AI tools. Anyway, perplexity is an innovative research tool that packs impressive features for in depth searches or data analysis. It has an advanced multi step reasoning process under the hood. You're familiar with GPT and Google Search. Using perplexity AI will be super easy for you. The core feature it offers is the pro search mode. It multiplies your searches and then dissects the most relevant information into a response. And what's the cool thing is that you also get the sources. The P search would analyze the search results, take intelligent actions and initiate follow up searches that built on previous findings. Perplexity offers a free version and a paid version with additional features for around 20 bucks a month, which seems to be industry standard at the moment. With the paid version, you can integrate almost any of the high end models currently available like CID GPDD, for latest version of Cloud model, Gemini, without the need to buy separate subscription for each service. You can also attach files such as PDFs, images or text files directly into your queries. Perplexity also features a quick search, and it's ideal for fast, accurate answers, backed by reliable sources. However, for in depth research, requiring comprehensive analysis, upgraded pro search is way, way better. Free version limits pro searches, but it's enough to decide whether it's worth paying for or a free version is sufficient. The free version of perplexity also allows you to enter detailed custom instructions, like ajipti. But here's one more thing that you can do in Chagp, well, you can organize your searches into collections by topic or project. This way, you can easily find your past searches by topic, you can even share a collection with someone. One of my favorite features of perplexity AI is focus. It allows you to narrow down your search to specific areas like academic papers, read it, or Wafram, which is ideal for precision and reasoning. For example, if you need to make an activities interests and opinions marketing analysis. Just type in your request, choose pro search, and click focus, and let's choose Rdit. And you'll get the results from reddit discussions. What I like about Reddit is that, while these are just people talking. And yeah, it's cool to get that insight, and it takes a lot of time to go through all of these threads and links. But what if you need to search elsewhere? No problems. Add site and column. And then type the site that you need to search. Preferably just copy and base it. From the next step, perplexity AI clearly cites its sources, making it easy to see where the information comes from, and it's even easier to verify this information this way. You don't have to search it manually. I also encourage you to pay attention to the follow up questions because they often are examples of a good prompt and a good way to follow up. So you can choose how you develop your threat without typing new prompt every single time. If for some reason, perplexity doesn't work for you and you're looking for a similar alternative, there's a direct competitor, and it's called. With, the approach is absolutely the same. Super easy to get used to and to impossible to resist using. Feel free to put the video on pause and read the conversations or even better, try them yourself in your context. Meanwhile, bye bye, and see in the next lecture. Oh 48. ChatGPT Free Alternative: HuggingFace Chat: In this video, I want to show you a cool alternative to the paid versions of HGPT, Google Gemini or EnthropiCud. It's called Hugging Face hat. Hugging Face, if you don't know, is a huge community dedicated to AI, particularly to the opensource. They have this hugging face chat that uses open source models. To use this, you'll need to create a hug inface account and go to hugface dot c slash CHAT. Once there, you'll see a familiar interface with similar to HA JPT features. Let's look at a couple of the components. When you start a new chat, you can choose a model. Let's check out the settings and select a model. Because this is free, sometimes the biggest and best models might be busy. If that happens, feel free to switch to a different model based on the current leaderboard. Metasama is one of the coolest open source models. You can also create your own assistant or use the ones created by the community. Once you select a model, you'll see a button called tools. Among the tools, there's image generation, image editor, URL fetch feature that you can use to parse URLs. Document parsing, calculator, and web search. Keep in mind that these are different tools from different sources. Most of them are open source, but they are of really great quality these days, especially considering that you can use them for free. I'm not sure how long it's going to stay free, because it obviously cost them millions or even billions to run, at least in terms of instance and computation. Let's base our chain of thought prompt as an example to compare. Now let's take a moment and compare how GPT four responded to exactly the same prompt. So please pass the course if you need to just to read and have a look at the comparison. In my opinion, there isn't much difference in the response quality. It's not that drastic and can be a matter of taste. I think they both followed my prompt pretty accurately. You can look at the assistance based on certain models or create your own. When creating your assistant, you can set up features like name, description, start messages, like conversation starters. The functionality is pretty similar to ha JBT. However, you can choose the model and add settings like temperature. By the way, temperature setting is very important in large language models. It defines how predictable or creative your response should be. With a maximum value of one, which is going to be very creative and zero, which is going to be very predictable. So depending on the task that you're trying to accomplish, you might want to adjust this setting. By the way, really quick update. Hugging Face has just added a ton of useful tools into custom assistants. They added a search of tools because there are so many of them now. You can just start typing the tool that you want to see. For example, fetch URL, which would allow you to get any link as a contact. Now really worth checking out, even if you're using a paid version of JGB. Now, let's go back to reality and talk about one of the downsides. There aren't many privacy settings, obviously, and you can't make assistance private, which means everyone can use them once you create one. If that's not critical to you, I'd say this will be a decent alternative to paying for HAGPT for many, many users. And if that's you, well, we just saved $20 a month, which is roughly $240 a year. Pretty cool, right. Hope this was helpful for you and see you in a couple of seconds in the next video.