Building Basic AI Agents

1. Building Basic AI Agents Intro: Welcome to building basic AI agents. Artificial intelligence is changing quickly. Not long ago, most people thought of AI as a tool that could answer questions, write text, or generate images. Today, AI is becoming something much more powerful, a practical partner that can help you plan, research, create, analyze, communicate, and automate real work. In this course, we are going to explore that shift step by step. We will begin by understanding the modern AI landscape and what makes AI agents different from traditional chatbots. A chat board usually waits for your instruction and gives you a response. An AI agent can go further. It can understand the goal, break it into steps, use tools, and take action on your behalf. This is the foundation for everything we will build throughout the course. Next, we will focus on one of the most important AI skills, communication. You will learn how to write clear and effective prompts, why some instructions produce weak results and how to use structure, context constraints, and examples to guide AI systems more reliably. From there, we will move into building your first custom AI assistant. You will learn how to identify tasks that are good candidates for automation and you will create specialized assistance for practical use cases such as video content optimization, image prompt generation, customer support and appointment booking. As the course continues, we will compare major assistant platforms, including custom GPTs, cloud projects, Jams, and autonomous tools like Mans AI. You will learn when to use each platform, how to build multipurpose assistant and how autonomous agents can complete more complex tasks with less manual input. We'll also explore advanced research and data intelligence. This includes using AI to collect information, analyze data, work with spreadsheets, automated emails, and even build simple web applications without writing code. Finally, we will bring everything together by learning how to design complex workflows. Instead of thinking only about single tasks, you will learn how to build systems connected processes where multiple tools and agents work together to save time, reduce repetitive work, and support better decision making. By the end of this course, you will understand not only how to use AI tools, but how to think with them, design with them, and delegate meaningful work to them. Let's begin the course now. 2. Understanding AI Landscape Today: Welcome to understanding the AI landscape today. In this lesson, we're going to give you a clear map of today's AI world. By the end of the next few minutes, you'll understand which AI platforms matter for your business, what's driving adoption across industries, and most importantly, where the real opportunities lie for you and your team. Let's dive in. Here's the reality. AI is growing faster than any technology in history. We're talking about a 36.6% annual growth rate. To put that in perspective, that's faster than the Internet grew in its early years. The potential economic impact could reach $15.7 trillion globally. That's more than the combined economic output of China and India today. Here's what matters for your business. 72% of organizations are already using AI in at least one business function. If you're not exploring it, your competitors are this isn't a trend anymore. It's a transformation. Now let's talk about the platforms. You don't need to master all of them, but you should know the major players. First, there's Open AIS chat GPT. It has over 200 million weekly users and is the enterprise leader right now. Then there's AnthropisCloud, which is known for advanced reasoning and can understand complex documents with a 200,000 token memory window. Google's Gemini is built into Gmail docs and Android, so billions of people already have access to it. And finally, there's Metasama which is open source and free to deploy. Each platform has different strengths, so the right choice depends on your specific needs. Here's the game changer. AI is evolving from something you talk to into something that works for you. Today, you ask hat GPT a question and you get an answer. That's a tool, but tomorrow you'll tell an AI agent to manage your customer follow ups, analyze your sales data, or schedule your team's meetings, and it'll do it without you touching it again. This is agentic AI, and it's coming fast. Soon a third of all enterprise software will have AI agents built in. This is where the real productivity gains happen. This is the shift from passive tools to active partners. So where should you focus? The companies winning with AI aren't using it for everything. They are picking a few high impact areas. Think about your biggest pain points. Are you struggling to forecast demand accurately? Do you want to personalize customer experience at scale? Are your team members drowning in routine tasks that could be automated? AI agents excel at all of these. The key is picking one or two workflows where AI can deliver wholesale transformation, not just incremental improvement. That's where you'll see real RI. That's where the magic happens. Here's what you do next. First, identify one high value workflow where AI could save time or unlock growth. Second, experiment with a platform trhatGPT Cloud or Gemini. Start with the free version, spend an hour exploring, see what they can do. And third, learn the basics of prompting. The better your instructions, the better your results. You don't need to overhaul your entire business tomorrow. Start small, pick one workflow, experiment, learn, and once you see the value, you'll find more places to apply it. That's how you start your AI journey, and the time to start is now. 3. From Passive Tools to Active Agents: Today we are talking about a massive shift happening in how we work with technology. The move from passive tools to active agents. Stop just asking AI for help. It's time to start having AI actually work for you. Let's dive into how this is changing business right now. Let's start with what we know passive tools. Think of Chet GPT or traditional software. You are completely in control, but you are also completely in charge. It's a pull model. You have to ask the right questions and every conversation starts fresh. Result, you stay busy managing the AI instead of managing your business. Now, meet the active agent. This is where AI takes action on your behalf. It works autonomously, 247. This is a push model. Agents identify opportunities and execute tasks without you lifting a finger. They remember context, learn from experience, and handle things like autonomous research and scheduling while you sleep. Why does this matter? The numbers don't lie. Research shows that using the exact same AI model, an agent workflow can achieve 95% accuracy compared to just 48% as a standalone tool. Agents can work on unlimited tasks simultaneously, eliminating repetitive errors and freeing your team to focus on strategy. So what does this look like in the real world? Here are five ways to automate your business today. Project tracking agents that alert you to delays, lead qualification agents that schedule follow ups, email management agents that prioritize and draft responses, data inside agents that flag anomalies and scheduling agents that handle your calendar. It's time to move from passive to active. Your action plan is simple. Identify just one repetitive task in your workflow and ask. Could an agent do this better? Start small and automate one process this month. The competitive advantage goes to early adopters and your future self will thank you. 4. Evaluating Modern AI Platforms: This video, we will talk about the main competitors clot has. If we talk about Cloud, definitely, it's a great AI tool that has a lot of possibilities, especially in the pro version. Definitely, we can use cloud projects, we can use Cloud attributes and it gives us possibility to write text and analyze data, create code, et cetera. But there are a lot of other powerful competitors and in this video, we will share how different AI tools can be beneficial for you in solving different tasks, solving different issues. The first one definitely we will start from HTGPT by Open AI. HGPT as Cloud has its own paid versions and free version. Free version is pretty limited and has a lot of small issues. You can't have a lot of requests and a lot of really prominent functions are not accessible, the same as in the cloud free version versus cloud pro. So if you talk about the basic h GPT models, you have hA GPT O and h GPT 4.5 at the main models we are working with. Also you have reasoning models that can help you to go in more deep reasoning, understand how to build your strategy, how to conduct research, et cetera, and if we talk about the most powerful GGPT models, they are one and all mini and all three high. Definitely, you can find more models here, but that's it as basis. If you talk about the main functionality of HGPT, it can go online. It can go and make you deep research. For example, right now, I was just conducting the research about SO strategies and here we are, we see that they were scanned 30 different resources. It was taken 10 minutes before HGPT gathered all these data. We can go through the resources, we can check it, et cetera This is definitely a great feature that we can use in HGPT. GPT can generate images. In paid version, you have Embded the LE and with the LE, you can generate any images you need and it's pretty useful and it's pretty easy. Definitely, all these benefits are not so crucial in HGPT compared with the huge army of small AI systems which called MGPT in cloud you already familiar with projects and projects when you can use your proms when you can feed, cloud your data. In Chant GPT, definitely, it's the same option where you can go to M GPTs and on the one hand, you can use your on my GPTs, but there are a lot of already created GPT. Here you can find image generator, writer, video, AI, logo generator, et cetera. You have here a huge amount of GPTs for different tasks of writing productivity, research analysis, education, lifestyle programming, et cetera, and this is amazing. We can go deep and we can create our own GPT. Just click on my GPT, you have two main ways, how you can do it. You can create it in chat. For example, if you need your own copywriter, you can ask as copywriter, for example. Copywriter and it will create your basic version of my GPT, which you can further configure. Right now it's utilizing it. You can go to the configure tab and here you can add additional materials, you can add your knowledge base, et cetera, and it will be your personal assistant for different tasks. I definitely heavily use this feature. I love it. I have my own uh, GPT which help me with course creation with image generation, with another stuff and we can easily use this feature indefinite really different stuff. For example, I love to generate linked in posse. It helps me to learn Portuguese. It helps me to prepare to my triathlon, iron man, et cetera. Chan GPT, my DPTs are really beneficial. After you will go through all this stuff, it's definitely the first stage where you can implement it. But going deep, you can easily make a lot of personal assistance in different stuff. Example, one of my favorite my GPD is my digital marketing strategy helper. You're already familiar with this prom that helps me to create digital marketing strategies really fast and just in here website, I have all quick research with mission statement, unique selling proposition. I have here competitors research based on the open similar web or Sarah data. All the stuff in one place and it's really incredible and really amazing. That's what I really love in HGPT. You've already know what is cloud projects and you see the similarity of these functions. They are similar but at the same time different. Please play with my GPTs also share your experience, which one you love more, what is better for you, et cetera. The next big tool I want to talk in this video is. Gemini by Google is really amazing tool. It has its own advantages and disadvantages. But if you talk about the basic functionality, it's definitely pretty good in a lot of stuff. For example, it also has its own deep research. It has a canvas for docs and code writing. It generates much better photos. For example, if I will ask Jamini to create a photo, it will be photo realistic because image option is much better than Deli. If we talk about image creation and free image creation, this cat is definitely better than we've had just in Dali. If we talk about the main core features I love in Gemini compared to HGPT, this is about connection with Google services. If you are Power Google user, you use Gmail, for example, or Google Docs or any other Google stuff, you can integrate Gemini with your email and it can be your personal email assistant. I will help you to write emails, summarize emails, to get insights, and it's really amazing compared if we talk about productivity. We will get really amazing insights about how you can interact with all this stuff one by one. Another thing that I love in Germany, it's definitely about activation, different YouTube task. For example, you can ask it for generate different YouTube video summaries and it will do it really fast. You can integrate, as I mentioned, it to your email and you will get the summary of the last emails with summary table, you can check it, you can interact with it, that's it. I definitely love Gemini in everything, but it is Google oriented and Google connected. I image for image generation, well to model for video generation and Gemini is definitely pretty powerful competitor both for Clot and for TGPT. These three services in Silicon Valley, they often called holy trinity. For example, you can hear that marketers say if they have some complicated prompt, if they have some challenges, they can ask both all three systems to give them answer and they try to compare it, and this is another challenge. In the era, most of the competitors didn't have deep research CHAGPT didn't have deep research, when Gemini didn't have this feature perplexity was number one in everything that we need for a search. For example, if we want to get some really valuable data, we can find pretty fast data. We can make research, we can get pretty valuable information and that's what is really crucial in perplexity. When you want to get real time insights with precise data that you can believe that you can work with, et cetera. That's where we can go through, for example, if you are creating content and you use free version of HGPT, it's going be laggy, it can hallucinate, et cetera. In the case of the perplexity, if you want to use perplexity for content generation, you first get proven data and after that, it's transformed in content, which is really amazing. We are moving to the next big thing is Dipsk. Dipsks definitely game changer because its tokens are much more price savvy compared to other competitors. But as Chinese AI, it has a lot of people who don't want to give personal data here. Dipsk give for free in open source model many things which were prohibited previously. Right now it's something they asked me about my edge. You will know my date of birth. You will maybe I will also Okay. And this is what in Deep Seek really amazing, it's deepsk R one, which is reasoning model, and you can get your strategy, and you can get a lot of valuable insights here. With search model, you can get and research and reasoning all in one and it's really amazing. There are a lot of really valuable stuff in Deep Seek, but usually I use it for reason so in the case, I need to get some stuff to better reasoning model. Definitely, I go to Deep seek because I can watch everything, how this model works, how it helped me to understand target audience, how it's reasoning, all the stuff. So it's a really big thing which I really love. For example, I will show you how it looks like. When you start chatting with Di Disk ARO, you have all the argumentation, so I was thinking for 80 seconds, why it's working like this, why, the budget was allocated to this channel, all calculations. You can double check this calculation. Dip Sik is amazing, but there are some small issues with questions connected with China, for example, when we ask about China and Tibet, China and Taiwan, you will get the answers which are more about let's talk about something else. The last but not the least in this great cohort of tools is Rocket the part of XIIon Mass company, they first gave available to all customers deep search model, so you can go and get a lot of valuable insights. You can see my researches, for example, I was doing research for SLL companies. You can see how deep they are going in profiles, how it interacts with different challenges. I will find some research which can be relevant for this stuff. You see how deep it is going on, so it's really amazing. Grock definitely can be valuable if you are searching for some fresh news because it's connected to X. In the case, you want to find something really actual, for example, what President Trump was telling about the President Zelensky, you will get not just summary of old news, you will get super fresh news from X and this another advantage from rock. So as you see, there are a lot of amazing stuff that you have in one place and definitely you need to test grog before you will decide to do something else. Guys, I hope this quick review was useful for you. Cloud is definitely amazing. I love this too. We use Cloud a lot in our daily routine. But Chet GPT, Gemini, perplex, T DPC and grog they have also their powerful feature that I recommend to test. 5. The Psychology of Effective Communication with AI: Welcome to the psychology of effective communication with AI. Here's the truth. AI isn't unpredictable. The problem is how we talk to it. In the next 3 minutes, you'll learn why some prompts produce gold and others produce garbage. Let's dive in. AI systems are brilliant, but new employees. They lack your context. They can't read between the lines. Think of it this way. If you showed your prompt to a colleague who knows nothing about your project, would they be confused? If yes, the AI will be too. That's your golden rule. Here are the seven most common reasons AI prompts fail. Ambiguity. You say, make it better without defining what better means. Vagueness. You throw ten tasks into one prompt, under explaining. You give the AI less detail than you'd give a colleague. Lack of context missing constraints, unspecified format, and information gaps. Fix these seven things and your prompts transform. Care, the framework that fixes almost every prom problem. C is context, give background and motivation. A is action. Be crystal clear what you want. R is rules. Specify boundaries and constraints. E is example, show the AI what you want. When you use care, you eliminate ambiguity and vagueness all at once. Three techniques multiply your results by 60 to 80%. First, specify output format. Tell the AI exactly how to structure results. Second, use examples, show the AI a sample of what you want. Third, ask for step by step reasoning. This forces the AI to show its work. These three techniques work with any AI system. Here's your action plan for the next week. Day one and two, write one regular task using Care. Day three and four, test it and compare results. Day five through seven, apply Care plus one technique to a second task. Then build a library of your best prompts. Clarity beats cleverness. Now go build something amazing. 6. Crafting Instructions That Actually Work: Welcome to crafting instructions that actually work. If you've ever asked an AI tool to do something and gotten a response that missed the mark, you're not alone. The difference between getting mediocre results and exceptional ones usually has nothing to do with the AI's capabilities. It comes down to how you structure your instructions. In the next few minutes, we're going to walk through a proven framework that business owners and project managers are using right now to get consistent, high quality results from AI every single time. By the end of this lesson, you'll have a practical system you can apply immediately to your work. Let's start with the problem. Most instructions fail because they lack structure. Think about it. When you say write about marketing, the AI has to guess what you actually want. Is it a blog post email, a social media thread for an audience of beginners or executives? In what tone? Without clear scaffolding, you get generic outputs that require heavy editing. But here's the good news. This is completely fixable. The difference between a vague instruction and a crystal clear one is structure, and structure is something you can learn and apply immediately. Meet the As framework. Ray stands for role, action, context, and expectations. This is the structure that transforms vague instructions into crystal clear directives that AI tools actually understand. R for role means telling the AI who it should be. Instead of just analyze this, you say, you are a senior financial analyst with 15 years of experience in corporate strategy. This activates the right knowledge patterns in the model. A four action is the specific thing you want done. Use strong verbs, analyze draft reflector, evaluate the bug, not do something with this code, but reflector this Python function to improve readability and reduce time complexity. C four context is all the background information the AI needs. Who's your audience? What are the constraints? What does success look like? This is where most people skip details and get mediocre results. A four expectations defines how you want the output formatted and what quality you need. Do you want JSON, bullet points, code examples? How technical should it be? Let's see as in action. Here's a real example. Without race, you might say, review this code, vague, right? With RAC, you'd say, you are a senior Python developer with expertise in API development and security. Review this flask endpoint for security vulnerabilities and performance issues. This handles user authentication for a financial application processing sensitive data. It gets 10,000 requests per hour during peak times. Give me findings in this format. Severity level. What's the issue? Line numbers, how to fix it with a code example and why it matters. Put security issues first. See the difference. One is a question. The other is a complete blueprint. Here's another example without race. Write about machine learning. With race. You are a technical writer for mid level web developers with three to five years of experience. Create a blog post outline explaining gradient descent. The audience needs practical understanding, not academic depth. Structure it in five to seven sections, include one real world code example, and keep technical jargon minimal. Specificity compounds. The more detailed and structured your instruction, the better the result. Now let's talk about the mistakes that kill your results. These are the patterns I see business owners and project managers making all the time. First mistake, assuming shared context. You know your business, your audience, your goals, but the AI doesn't needs you to spell it out. Don't assume it knows what professional tone means to your company. Second mistake, mixing instructions and context without clear separation. Use headers or delimiters. Put instructions at the top, then context. This helps the AI understand what's a directive versus background information. Third mistake, fluffy language. Make it fairly short is weaker than 200 to 300 words. Kind of like this is weaker than showing an example. Precision matters. Fourth mistake, only saying what not to do. Instead of don't use jargon, say simple language suitable for non technical stakeholders. Tell the AI what to do, not just what to avoid. Fifth mistake, overloading with irrelevant information. Yes, context is important, but too much context confuses the model. Give what's needed, nothing more. The fig use raise consistently. Test your instructions with sample data, refine based on results. Treat your instructions like a reusable asset that you can improve over time. Here's your action plan to start using RAS today. Step one, pick one task you do regularly that involves giving instructions to AI. Maybe it's writing marketing copy. Maybe it's analyzing data, maybe it's reviewing code. Pick something you do at least once a week. Step two, write that instruction using Rs. Define the role, state the action, provide context, set expectations, write it out completely. Step three, test it with sample data, run it through your AI tool of choice and compare the results to what you used to get. Step four, refine based on what worked and what didn't. If the output is still missing something, adjust your instruction. Add more context. Be more specific about expectations. Step five, save your best version as a template. You're building your personal instruction library. Over time, this becomes your competitive advantage. Your homework this week, apply as to one instruction, test it, document what works, build your library, the payoff consistent high quality results. Less time editing, more time on strategy. That's what we're after. Thank you for watching. Now go craft some instructions that actually work. 7. Proven Frameworks for Consistent Results: Come back. In this lesson, we're going to dive into the best practices for writing high quality prompts when working with Cloud AI or really any large language model. The way you phrase your prompt directly impacts the quality of the response you get. So if you want better outputs, it starts with better inputs. Let's begin with a simple idea. Be specific. The more clearly you describe what you want, the more likely you are to get it. For example, if you ask Claude, write something about marketing, you'll probably get a general overview. But if you ask, write a LinkedIn post promoting a new email marketing course targeted at small business owner now you've given the AI a direction, a purpose, and a tone to follow. That brings us to another key practice. Define the goal of your prompt. Ask yourself, what are you trying to achieve with this output? Are you informing, persuading, summarizing, creating, or analyzing? When you tell Claude what you wanted to do, it becomes much easier for the model to align its response with your intent. Third essential best practice is provide structure or format. If you want a list, say so. If you want bullet points, an essay a table or a social media caption, include that in the prompt. For instance, create a bulleted list of five tips for improving remote team communication is far more likely to give you exactly what you need than just saying tips for remote work. Now, let's talk about setting the tone. Claude adapts well to different styles, but only if you guide it. Use skills like writing a professional but conversational tone, or make this sound friendly and enthusiastic as if you're talking to a new customer. Think of tone as the voice behind the message. Setting it early helps ensure consistency. Another helpful technique is role assignment. When you ask Claude to take on a persona like a teacher, a product manager, a customer service rep, it frames the response in a more useful context. Try you are a project manager explaining this to a new intern and watch how that shapes the reply. Here's a big one. Test and refine. Don't expect the first prompt to be perfect. Writing great prompts is often an iterative process. You might need to rephrase, expand or adjust your wording to get better results. Small changes like shifting from an open ended request to a more focused question can make a huge difference, and finally, avoid ambiguity. Vague terms like it this or something interesting can confuse the model and lead to generic or off target answers. Try to write your prompts as if you're explaining something to a smart colleague who's helpful but doesn't read your mind. Delivered with your words. To wrap up, high quality prompts are specific, goal oriented, well structured and intentional about tone and draw. And like any good craft, prompt writing gets better with practice. Let's keep going in the next lesson. 8. Identifying Automation Opportunities: Welcome. In the next few minutes, we're going to walk through a proven framework for spotting automation opportunities in your business. Whether you're managing a team of five or 50, this approach will help you identify which tasks are wasting your team's time and energy and which ones you can automate to free up your people for higher value. We're going to show you exactly how to spot automation ready tasks, how to prioritize which ones to tackle first, and how to avoid the common mistakes that derail automation projects. By the end, you'll have a clear action plan to implement your first automation initiative. Let's get started. Let's start with the reality. Manual work is expensive, not just in terms of time, but in errors, frustration and missed opportunities. Here are the numbers. Organizations with substantial automation complete their financial close in six days or fewer. Organizations without automation, it takes them much longer, and that's just one example. When your team spends hours on data entry, scheduling or report generation, they're not thinking strategically about growing your business. They are stuck in the weeds, and here's the kicker. Manual processes are significantly more error prone than automated workflows. One wrong number in an invoice. One miss deadline in a project schedule, and suddenly you're dealing with customer complaints or compliance issues. The good news this is exactly where automation shine. Automation doesn't just save time, it improves accuracy, reduces stress, and frees your team to focus on what they do best. Here's the framework you need to remember. When you're looking at your workflows, ask yourself these four questions. First, is this task repetitive and time consuming? If your team does it regularly and it eats up hours, that's a red flag in a good way. Think data entry, report generation, file organization, these are prime candidates. Second, does this task create errors? Manual data entry, copying information between systems? That's where mistakes happen. If you're constantly fixing mistakes or dealing with duplicate entries, automation can solve that. Third, does it follow the same pattern every time? If you could describe the steps to someone and they do it the same way every single time it's automation ready, tasks like sending reminders, responding to common inquiries or routing approvals, these are perfect. And fourth, is this step holding up the rest of the process? If approvals or manual checks are bottlenecking your workflow, automation can be a game changer. It keeps things moving. Now here's where a lot of businesses go wrong. They try to automate everything at once. Instead, think strategically. Look for quick wins first. Quick wins are tasks that happen frequently are simple to automate and will free up real time for your team. A quick win builds momentum and confidence. Once you've got one automation running smoothly, you can tackle bigger, more complex processes. After your quick wins, move to strategic priorities. These have higher impact but require more effort. Think invoice processing, financial reporting or HR on boarding. These are worth the investment because they save significant time or prevent costly errors. Here's a critical insight. Don't automate a broken process. If your approval workflow is chaotic or your data entry process is messy. Fix that first, then automate it. Automation amplifies efficiency, but it also amplifies problems. And remember the 80 20 rule. Most of your time savings will come from automating a small number of high volume tasks. Focus there first. Let me share some hard won wisdom from businesses that have been through this. There are five mistakes that derail automation projects. First, starting without a plan. You need to know what you're trying to achieve. Are you saving time, reducing errors, improving customer experience? Be specific. Define what success looks like before you start. Second, unclear roles and responsibilities. Your team needs to know who's setting up the automation, who's training everyone and who's monitoring it. Unclear ownership leads to chaos. Third, skipping staff training. Your tools are powerful only if people know how to use them. Invest time in training. I pays off in adoption and results. Fourth, no backup plan. Technology fails sometimes. You need to know how to keep the business running if automation goes down. Document your manual workarounds and fifth, ignoring measurement. You can't improve what you don't measure. Track your time savings, error rates, and productivity improvements. This data will help you justify automation investments and identify your next opportunities. Alright, let's make this real. Here's what you're going to do this week. Sit down and list the tasks your team does regularly. Which ones are repetitive? Which ones create errors? Which ones are holding up other? Just list them. Don't overthink it. Next week, pick one. Your first automation should be something that happens frequently is relatively straightforward and will clearly save time or reduce errors. Define what success looks like. Maybe it's saving 5 hours a week or cutting invoice processing errors by 50%. Then research your options. There are tools out there for almost every type of automation. Within 30 days, implement it, train your team and measure the results. And here's your homework. Document one workflow from start to finish. Write down every step, every decision point, every approval, share it with your team and talk about where automation could fit. This is how you build a culture of continuous improvement. You've got this. Let's automate your way to a smarter more efficient business. 9. The Specialist Assistant Video Content Optimization: In this video, we are talking about how to use JTPT or any other LLM to enhance your prompts in AI video creating services. In different systems, you have already this option to enhance prompts with AI, and if you will start with some basic prompt, for example, the more shooting selfie, you definitely get different output and predictable output, et cetera. How we can enhance the prompt inside, for example, GPT. You can ask help me to enhance the prompt. Help me to enhance the prompt, prompt for AI video creation. Make the video professional. Professional. You will get some prompt enhancement. Here we have a high resolution cinematic video of a curious ring tailed lemur in a large Madagascar forest, holding a smartphone and playfully taking selfie. With all details, et cetera, here you have additional options to control. If you want to you can add to prompt camera type angle framing, style reference, lighting, et cetera. Even on this level, we already have the t high enhancement of the prompt. Just some basic details, we already have prompt much better than our lemur shirt in self. What you can do further. If you have your patterns, which really works, which give you great results in your AI video generation, you can go to MGPTs and try to find here some prompt video generator. And depends on your task, you will get, for example, so video prompter or clink image to video prompt helper or cling AI prompt. So it depends on the system where you want to create it, you definitely will get pretty good solution even here. But if you have your own experience, you have your own examples. I highly recommend to create your own MGPT that will help you to enhance the prompts and get really grate photorealistic results. Because in the case, you will create your own M GPT. It's possible just in paid account of the GPT. It's not available in free version. You have the possibility to teach M GPT and give your own prompt examples, your own settings, tone of voice, brand identity, et cetera, it can be done in one place. Let's go and let's create. So we're clicking on the create button, and here we have the chat where you need to start. We have help with AIVDaPmpt enhancer. So let's see how it will work for our case. Even with the basic setup, you will get some kind of installing, but definitely you need to add more details during the setup. Cinematic prompt crafter, would you like to go is okay, I like it. So it will create details. It's enhanced prompts VYI video tools with cinematic detail and creative direction. So let's go further. While it's creating image, yeah, we have right now image, we can go to configure and we see that this GPT acts a creative technical partner for enhancing prompts intended for AI video generation tools like Rhyming Wave PC, et cetera, so it can be any of the tool that we need. But the most important is to go here and upload your files. If you already have your own library of proms which works well, you can edit. If you have your tone of voice, your brand identity, any details that can be helpful during the crafting of the prompt, I highly recommend to go here and upload this information. What is the next stage? It's to create this cinematic prompt crafter. You can make it accessible just for you. You can share with your colleagues by link. So let's just make accessible just for you. And right now, we have our GPT, which will also available here in the left menu. And if you will try to play with Lemur shooting Selffi we can get it here. So we can get it here. So we have detailed pretty full prompt, and we can use it in for the approach. We can add any additional details GPT is asking, do we need any details? So you can do it not just in DBT. You can try to play it in clot projects, in Gemini extensions, et cetera. So it's pretty straightforward in any LL lambs that you use, but I highly recommend you to enhance the prompts. You can inhale the proms with other proms, just asking LML lamb what results you want to get. But if we talk about the next big stage, you definitely have the possibility to implement all this knowledge in your own M DPT, and it will be your powerful system that will make your life much more easier, your video production much faster and more efficient. So try it, see how it works, test it, and I will see you in the next video. 10. The Creative Partner AI Image Generation Assistant: In this video, we will talk how you can use GPT to enhance your prompt or create even your own MTPT to make your prompts better for AI image creation. You can use these prompts in any AI image generator folder or even can use in Build ChatGPT Dali image generator to get astonishing AI images. So let's start with the basic stuff. For example, if you want to generate some baby SOT, some cute image that you will post in your social media. You can go to the create Image option and start generating with GPT. Definitely, it's not the best way how you can do it. GPT will create some image, but it will not be so good as you will enhance the prompt previously. There are a few ways how you can enhance the prompt. The first way is to dig Internet, find the best prompt libraries, go to find some GPT Bibles and try to write big astonishing prompt. And we have in this course lesson where we talked how you can do it. It's the first way how you can go through to get the basic understanding what do you need to change in your prompt? It's about role, it's about the contents, it's about format, the first way how you can implement it. You can already see that GGPT is generating something, but definitely we will enhance the prompt previously, it can be better results and we can generate much cute photo realistic and well organized photo than in the previous case. What is our approach in this case? The first and the most basic way how you can get better prompt. Enhanced prompt is to go to ChatGPT and ask it to create a better prompt from your basic one. Definitely, baby floth is a bad example of our prompt. It's definitely not one we need to use for our approach in EI image generation. We need to go to the next stage and we need to enhance our prompts. Let's see how it can work. We are going to ChatGPT and the basic prompt for the prompt enhancement and can be helping to enhance the prompt to create photo realistic baby lows. You can change baby flows for any detailed task or image you want to create. We will go to the ChatGPT and we will get this recommendation. We have pretty detailed prompt for photorealistic baby lots, cleanings, tree branch, a large rainforest, the quality of national geographics, and if it's needed, we can get some details for style, for example, wildlife photography, camera, sat, lighting, et cetera. If you have some detailed case, for example, you want to use, for example, Mid journey, you can ask to get your recommendations in this prompt exactly for mid journey and you will get these details in this. This is the first way. The second way is to create your own promised enhancer. For example, I have this one for my Ukrainian audience, and for example, I want to do the same. I'm just pushing here my short prompt. I want to create baby slot, and based on my goals, it will give me the basic questions. I need to go through and I need to enter my GPT. For example, I want this baby slot in jungle smiling is a good idea. I want to be photorealistic. And I want to do it, for example, for Mid journey or any other AI tool. And based on the data that I've shared here, we have the detailed prompt. We can go to the Mid journey and use image common to generate this prompt and make and enhance it. What is really possible for everyone depends on what system you are going to use. You can create such Photopmptmster, your own photopmpt enhancer and make your life much easier. A lot of systems already have this prompt enhancement, so you need just to click and you will spend some credits to get better prompt. But my recommendation is to go to ChatGPT. You can try to search some ready to use photoprompt enhancer care. And we have here journey photo realistic image prompts, photo prompts, flux la professional proms, et cetera. You can use any one of them, or if you like exact AI image creation tool, you can just go to create and based on the chat, you can give here details. I need my GPT, which will enhance prompts, my prompts and generate photorealistic images. If you need any other styles, for example, anime, et cetera, you need to do it and here you need to write for which AI image generation system do you need it. After that, it will go starting creating GPT GPT for this goal. And you will see how fast it can do it. So our next stage it's to set up the icon. It will do it automatically. So it will generate the icon for this prompt generator. It will give the name for the real prompt enhancer. We can say yes. We can say that it's okay. I will create profile picture. While it's creating profile picture, the most and the most important settings are here in the knowledge base. If you already have a library of well working prompts, I highly recommend to upload here files with these prompts with details, you know that they work. This is your treasure. This is your AI goal because if you will give this data to GPT, HGPT will work with this pretty good one. And so we have right now your GPT it can be only for and, for example, it's okay. And we can experiment with this again. We can also give here data for baby slots, and we will get some prompt. But just to recall, you need to go to your basic stuff. You need to add prompts from your library that you've already used, and they work best. And this will be a game changer in your experience. So here we have not basic prompt like baby flows. We have detailed prompt with our structure that we used previously, so tested. I share your experience in the QNI section, and I will see you in the next video. 11. The Support Agent Customer Service Automation: Hi, guys. Welcome to this lesson. In this lesson, we will go through how you can use 11 labs for customer support, and it's amazing feature, conversational AI. So let's get into it. So when you go to your homepage, here you have an option AI agent. You can also find it here in the left bar menu, conversational AI. And when you click on it, you see your number of calls, average duration of these calls, total cost of these calls, and average cost credits per call and some metrics. But because now we don't have any AI agents active, we don't have this data. But when you will have it, you will have the statistics that you can analyze and work with. So when you go to Agents, you can create new agents. So here I tried some agents before, but we will play and create a new agent. So click on New Agent, and here you can find a template and name it. So, for example, if we're using it for customer support, we will name it customer support agent, and I will choose this template already. But you can also choose the blank template and customize it to your needs. But because we're talking about customer support, we will choose a template that's available and click Create Agent. Then you will go to the settings thing, and here we have a really big number of things that we can tweak and how can we adjust it? So first is agent language. For us, it's English but you can also add any additional language. For example, we can add Polish, Ukrainian, and Spanish. So here will be all the languages that your agent will understand and speak. Then the first message, this is the first message that our agent will say, Hey, there, I'm Alexis from 11 Lab Supports. How can I help you today? You can write not 11 labs, but, for example, skills booster. Because it's our agent, you can also add variable and customize it and translate to all languages that are supported. Also, system prompt, this is the personality of your agent. You can also tweak it. You can use Chat GPT to make it more customized for your business, but we will not do it right now because we're already using the template that is okay for us. Then you use the LLM. Here you can use anything that you like, and I will leave what it was Gemini two point Flash. Then you control the temperature. It's a parameter that controls the creativity or randomness of the responses generated by the LLM. Here, it's 0.5 z. You can make it bigger, you can make it smaller, but I will leave it as it was. Limit token usage, configure the maximum number of tokens that LLM can predict. I will leave it at minus one. Here you can also add a knowledge base. So here will be a document that your agent will use when answering questions. But for now, because we're using this as example, I don't have any documents, but you can add URL, you can add files, or you can just write text. Then you can add tools. So for example, it has end call so that the user can end the call and you can add some other tools here. Save. Next, we go to voice. So here's the voice that our agent will use. You can use even your voice or any voice that you want. We will stay with Jessica. And here you can add pronunciation dictionaries also stability, speed, similarity, like we did before, analysis, so you can add evaluation criteria and data collection, something that the agent can collect from your calls security. So you can also add allow lists, enable the overrides and stuff like that, and advanced. Turn timeout, silence and call time out because we're using the template. These are already ready to go settings, but you can read everything here, and the 11 labs provides explanation for each setting and put the amount that you need. You can even ask Chat GPT on how to use it. And next comes widgets. So how it will look like. We can also We can also edit the colors, and I want to edit the colors of the avatar. I have my brand book here and I can change the colors of the avatar of our agent and make it like this. You can make it orb or add a link or image, whatever you like. Text context. What buttons will it have? Start a call end the call, need help, listening, talk to interrupt. And the interface, you can add a language drop down so that our user can use whatever language they understand, add terms and conditions and share all page, and this is it. Next, you click Test agent and talk to it just like you talk to ChadGPT or any other AI agent, and it will help your customers and you to answer your questions. You can also edit name, copy agent ID, delete this agent, or go to conversation history, and you can also copy link to this agent. You click Save. And now you can find your agent in Agents. Here is our customer support agent. 12. The Scheduler Appointment Management Agent: Hello, everyone. Welcome to today's lesson. We're diving into another fascinating case study, and I'm excited to guide you through it using 11 labs. As we've explored in our previous discussions, we've delved into conversational AI and the creation of AI agents. Today, we're focusing on a practical use case, crafting an AI agent specifically for booking appointments. When you access your main dashboard, take a look at the left side panel to find the conversational AI section. Within this section, you'll find the agents area where you can create a new agent. You can start with a blank template, but since I've already set up an agent, I'll walk you through what it looks like. The initial step involves selecting the primary language for your agent, along with any additional languages your AI might need to accommodate. For instance, if your services extend beyond English to languages like German, Ukrainian, and Polish, you would select these languages too. This leads us to setting up the first message. For the first message, I use ChatGPT to craft it, but feel free to tailor it to your needs. Here's how my agent appears, and I made sure to click on translate to all, so the message automatically adapts to any selected language. Next, we move on to the system prompt. You have the option to write the system prompt yourself or use ChatGPT like I did. You'll need to articulate what you want your agents to accomplish. For example, I requested ChatGPT to generate a system prompt for an 11 laps AI agent focused on booking appointments, and it produced a suitable prompt. Feel free to customize this prompt to your liking, modifying whatever you find necessary. This is where you establish your agents persona. Afterwards, select the LLM version that suits your requirements. I'll be sticking with what 11 labs offers for this. You can adjust parameters like temperature for creativity and set limits on token usage. Adding a knowledge base is also an option, though, since this is a conceptual case, I haven't added one. Adding a knowledge base is advisable, and you can input secrets or other tools here as well. Remember, it's crucial to save your changes to prevent losing any progress. Next, you'll select a voice for your agent. I went with Archer, but there are plenty of options provided by 11 labs or you can upload your own. Consider turning on the flash feature for low latency scenarios. Personally, I have it enabled. You can fine tune the output format, incorporate pronunciation dictionaries, and optimize for streaming latency. Adjusting factors like stability, speed, and similarity is also possible. I opted to slightly reduce the speed for more clarity. Now we look at analysis. Here, you have the opportunity to establish evaluation criteria or data collection methods. It involves simply naming and detailing the prompt. Moving on to security, you can enable authentication if desired, though I have chosen not to for this instance. Implementing allow lists to specify permitted hosts and enabling overrides for client requests are also available options. I have opted only for an agent language this time. Web hooks and daily call limits can also be configured if needed. In the advanced settings, you can adjust the turn time out, which is the maximum time allowed since the user last interacted. You can also set a silence and call timeout, which determines the maximum silence duration before a call terminates. Here you define the maximum conversation duration and input relevant keywords. I've included terms like book, reschedule, cancel, appointment, details, preferred time and schedule appointment to help the agent identify key topics in dialogues. You also have the option to specify the user output audio format and manage client events such as audio interruption, user transcript, and agent responses. Additionally, there's the option to save call audio and decide on conversation deletion timelines. In the widget section, embedding code and gathering feedback can be configured. You can choose when to request feedback. The feedback can be ignored, collected during the conversation or after it concludes, customizing the widget appearance is also possible. I've picked colors that appeal to me. The widget can feature a unique look like a round shape. I've added my preferred color codes, or you can opt for a more orb like design. Adding a link to an image such as an avatar is another option. However, be mindful of the recommended resolution and maximum size. For now, we're sticking with the orb design. You can also determine what will be displayed on the widget, such as start a call a call a N help button, listening status, and instructions for interaction. For the interface, I've enabled the language drop down so users can select their preferred language. While muting during calls is an option, I've left it disabled for now, including terms and conditions here is crucial for informing users about your policies, and there's also the option to add a sharable page. I've written a brief description of our AA agent, highlighting that it's an AI powered voice assistant designed to assist clients in booking, rescheduling, and canceling appointments for my enterprise. Once everything is set, I save these configurations and proceed to test the AI agent. You can select the language and interact with it to see how it handles frequently asked questions. 13. Comparing Assistant Platforms: Welcome to Skills Booster. I'm excited to walk you through one of the most important decisions you'll make this year, choosing the right AI assistant for your business. Right now, there are four major platforms competing for your attention and your time. GPT four Cloud Gemini, and Manus. Each one promises to boost productivity, but they work in completely different ways. By the end of this lesson, you'll understand exactly which platform solves which problems in your business. No fluff, no hype, just practical clarity. Let's dive in. Let me introduce you to the four platforms you need to know about. First, custom GPTs from OpenAI. Think of this as the creative generalist. It's incredibly easy to set up requires zero coding, and it's perfect for brainstorming, copywriting, and creating custom tools for your team. Second, cloud projects from Anthropic. This is the detail oriented specialist. If you need to analyze complex documents, debug code or do work that requires precision, Cloud is your go too. Third, Google Gemini. This is the speed demon. It has the largest brand in the room, literally the biggest context window, so it can process massive documents and datasets faster than anyone else. And finally, Minus AI, this is the game changer. Unlike the others, Minus doesn't just answer questions. I actually executes tasks. I plans, it acts, it delivers results independently. Here's the fundamental difference you need to understand. GPT, for cloud and Gemini are passive assistants. You ask them a question. They give you an answer, then you do the work. You copy, paste, edit, implement, it's a conversation. Manus is different. It's an active agent. You give it a task with clear instructions and it goes to work. I plans the steps, executes them, adapts if needed, and delivers the result. Not in a back and forth conversation, you're delegating. Think about the time difference. With a passive assistant, you might spend 20 minutes asking questions, refining answers, and implementing. With Manus, you spend 2 minutes giving instructions. Then you move on to something else. This is the future of productivity, not smarter conversations, but smarter delegation. So how do you choose? Here's the practical decision guide. If you need creative ideas, whether it's campaign concepts, social media copy or brainstorming, go with GPT four. It's the most versatile and it's incredibly easy to use. If you need precision and analysis like debugging code, reviewing contracts or working through complex logic, Claude is your specialist. The most accurate for detail oriented work. If you're drowning in data, huge research projects, massive document analysis, or real time information gathering, Gemini is unbeatable. Its context window is literally 1 million tokens. That's like reading an entire book in one go. And if you want to automate your workflows, email campaigns, appointment scheduling, data entry reporting, Manus is the only one that can truly execute independently. Most successful teams use two or three of these. The key is matching the platform to the problem. Let's talk about learning curves and time to value because those are the two things you care about most. Learning curve is different across platforms. GPT four and Cloude, you can start using them in 5 minutes. Gemini takes about 10 minutes to get comfortable. Minus takes a bit longer, about 30 minutes because you need to think about workflow automation differently. To your first win. That's where it gets interesting. With GPT four, you'll see value in your first prompt. With Claude maybe 15 minutes, Gemini 30 minutes, but manus, give it one to 2 hours of setup, and then you're saving hours every single week. The real question isn't the upfront cost, it's the time you save. Most teams report five to 10 hours saved per week once they've integrated these tools properly. Here's your action plan. Three simple steps to transform your productivity this week. Step one, identify one recurrent task that's eating your time. Maybe it's drafting emails, maybe it's analyzing reports. Maybe it's scheduling meetings. Pick one. Step two, choose the right platform for that task. Use the decision guide we covered. If it's creative work, go GPT four, if it's analysis, go Cloud. If it's data, go Gemini, if it's automation, go Manus. Step three, spend 30 minutes experimenting. Don't overthink it, start. Try a prompt, see what happens. Iterate, bonus step, document what works, share it with your team, build your playbook. This is how you scale productivity across your entire organization. First win is 30 minutes away. Let's go. Thank you for watching Skills Booster. Now go pick your platform and start saving time today. 14. Building Multi Purpose Assistants: Hey, there. In this video, we are talking about AI assistant. The basic AI assistant, they are based on your functional prom that you've created to solve some tasks, write the text, create the deck, to automate your sale script, et cetera. If you talk about the basic prompts, we've already talked about different frameworks, how you can structure prompt, how you can make it more efficient, how you can adopt these prompts to the next stage. And in this evolving world where everyone is moving to AI agents, the next things like my GPTs, Google Geminis gems and clot projects are examples of basic AI agents, but not autonomous. In each case, you need to implement and start using this stuff on the regular basis. You don't have your personal AI agent which starts automatically, but you can create your own AI system for different tasks for productivity, for marketing management, et cetera, to implement it on the regular basis. So if you talk about the idea of the AI assistant, you can create them in three main AI systems. It's GPT is opening AI, it's Google Gemini Gems and it's projects in clot. If you talk about the basic idea, what is the difference of M GPT, for example, and the systematic prompt. You have your own chat with your own database, we can say that it's not right, but the knowledge base that you put there and you start implementing this knowledge for further approach. Let's proceed and see how this works and how it can be implemented. The most powerful stuff of AI systems is in ChatGPT. You are just going to GPTs and here you have a full library of such tools. You have different tools for of writing, productivity, research analysis, education, lifestyle, programming, et cetera. You have the featured, you have the trending, you have the most popular, you have different integrations that you can implement further. There are a lot of different really valuable stuff that you can implement for them. If you talk about how it can be managed, if you need to create your personal AI system, for example, for writing content for your Instagram or Facebook, creating presentation, the first thing you need to try to do and you need to go to the GPTs library and start search and what do you need to have? For example, you start searching for Instagram helper, the GPT that will help you to create Instagram content. We have different GPTs. For example, we have Instagram post writer, Insta guide, Instagram expert, Instagram marketing expert, et cetera. We can click to any one of them, check what reviews do we have? Not so good 4.2, just how many conversation was done, and start the chat and test it how it works. For example, we can ask the write an Instagram post for this artwork. We have some example of Instagram post with hashtags, with understanding what is going on, et cetera. This is the bad example of assistance. Why? Because it's not well educated to make it working in the right way, you need to start educating this, you need to start giving a good knowledge base, a lot of information is not structured, so this is not the right way you can work with this. What is the right way working with M GPTs? You need to create your own. You're going to creating and here you have few more options, you have the possibility to configure it to write name, description, instructure conversational starters knowledge base, et cetera, and you have another more valuable options, you have the possibility to create it in chat. I'm usually creating in chat and after that, go to the configuration and add additional details. The start working with GPT is always starts from the systematic we can use any prompt framework that we've already learned in our course. For example, we can ask GPT to act as professional salesman, UD course creator, Instagram, marketing expert, et cetera. Any role we start with. After that, we are talking what we need to do, for example, act go through the process of creating Instagram account, writing scripts, preparing presentation, automating some reporting, et cetera. After that, we are given context. The context is the most important. We are going to the configured tab and here we have a plot files. For example, if you are writing content for Instagram, you need to upload your tone of voice, your target audience per solanas additional examples of your analytics, what works well and this is what is really crucial and very important. Let's proceed with creating, for example, Instagram helper, and we can start with Acts Instagram marketing expert. Marketing expert and the next stage, what do we need to do? Create a engage and post, engage and post on the topic from customer and give the text of post post image and hashtags and hash texts. You need also here hashtags. This is still very basic problem and very bad proms. Our M GPT will work 0-10, so it will not be so ideal. But we can push it further and see how it's going on with the configuration. We have the basic information and here in this prompt, we have role act as Instagram market and expert. We have the action, what we need to do, we need to create an engaging post on the topic from customer and we have the output that we need to get. We give the text of post image and hashtag. What is wrong with this stuff? We don't have enough context, we will need to add additional context here and in our case, we need to upload here target audience portrait, we will need to applaud here, additional stuff like more details about tone of voice. It's better to applaud your analytics from previous posts from the Instagram analytics, add additional details about which ports perform best, which post performed well, not so good and this is where we need to add additional details. Coming back here, we have our name we can say that it's our Insta marketing expert or marketing expert for Insta, GPT don't gives the possibility to share content words like such words as Instagram Facebook, et cetera, it's about trademarks. We have description, creates engaged in Instagram posts with text, images, and hash tags. We have the detailed description. We have the basic information that we need to do and as mentioned previously, we need to upload files. We can turn on the code interpreter to give us some table information. We can go through plot files to add the database. Here we can upload, for example, some tables with our previous Instagram statistic. We can add additional information, for example, about what is going on with our activities. We can some details about what is going on wrong and what is going on. The last but not the least in this story, we can go through the next stage in this activity. Is to try test it and see how it works. We are going to the creatab here we have three main options, what we can implement, it's only me, we have the possibility to don't share it and use just by ourselves. It's very important when you are using something privately, you don't want to share your knowledge base with somebody. Anyone with the link and you have the possibility to share with the team. To create My GPTs, you need the paid version of my GPT, at least the basic one for 20 bucks, but you can share it even with colleagues who don't use paid version, they can start using it for free. You have GPT store, we've already checked some of them which were ChatGPT store, you can easily go through and use it with anyone with. So we have anyone with the link. Let's see if it will be published because we have Instagram a lot of in the text and maybe it can be an issue and in one of work smooth. After that, we have our marketing expert for Insta, and we can ask, for example, to write a post about top ten AI tools for marketers, tools for marketers. Let's see how it will manage this. Here is what is going on. Here gives the Instagram post, top AI tools for every market needed in 2025 GPT definitely how we can live with GPT Just per sulfur, lumen, Copy EI, grammar go, Canva, notion, AI, Csa and Loco. We have this list. We have hash tag, we have the visual concept. You can ask create picture. And we will get the picture for this post it's still creating and based on our imagination, we have the visual concepts that we need to show the carusel and after that, it will implement it for them. In my case, I heavily use my GPTs. I have hundreds of them for different tasks and my issue that it's really hard to search them because GPT still don't have a well designed search tool for GPTs I have my American bro, which helps me with my English writing. I have my LinkedIn in helper, which helps me with LinkedIn in content creation, marketing strategy, other stuff, and this is really good when you have all the possibility to implement in your further approach. Another stuff which I use GPT regularly, I use my GPT. Make one AI to talk to another AI. For example, I have my DPTs who helped me to create and craft prompts for presentations. For Gamma app, I have my GPT which helps me to create images, prompts for open art, video prompts for cleaning, et cetera, and this where we have really high productivity and possibility. Definitely, this image will not be so good because in prompt, there are a lot of small details and the as you see, there are misspellings in the prompt. But if you ask just give us logos and don't add anything else, it will be okay to work further and implement. How you can make your AI systems better is educate them. You can go in any moment to the di GPTs and here you have the possibility to add additional details. You have here the full scope of details where you can upload files, upload your tone of voice. For example, if it's sales assistant, your scripts, cerem analytics, et cetera. Testd a similar functionality is available in Google mni's called Gems. You can go to the Gemini here, you are going to explore gems in gems you have pre made by Google. You have a lot of gems the same as library and GPTs library of GPTs. You have Chessam, you have Bred store Mark area guide, code partner, learning coach, writing an editor, for example, we are going to write an editor and here we have some ideas, what we can write. If you talk about Gemini process creation, it's pretty similar to other gems, you are going to explore gems. You create new gem and here you can create any helper which you need. For example, I want to create detailed gem let's copy some crater that I've already used. We can just copy paste prompts that we've already used in our further approach. Let me hide this from sidebar. Our marketing expert from Insta, and let's edit it. We will use the systematic prompt that we've already used in GPT and it's also possible. If you're operating in different systems, you can copy your system prompts from one to another, here we have the systematic prompt here and let's call it our Insta Bro. We have our Instagram quantat creator and it gives you possibility to work with it. We have Instea and we can start chatting with it and we can play the same game that we've already done. We can ask to write the post about top ten AI tools for marketers. We can go there and ask to write AI tools for marketers and this is okay. Here we have this data and you see that it works pretty well. We have the top AI tools with the same even number of tools, but a little bit different. It gives us the visual concept that we last but not least in this cohort, it's the clots projects, so you can educate the same. For example, this is my example of cloud projects that is educated to create digital marketing strategy from scratch and we have comprehensive data with all competitors, target audience, persona, example of content, et cetera, and it also was done on the same principle. As you see, it's pretty straightforward, you can to test to try and see how it works for you. I hope it will help you 15. Introduction to Autonomous Agents: In this lesson, you'll discover how autonomous agents are transforming the way businesses operate. Unlike traditional AI tools that wait for your questions, autonomous agents actively work for you, planning tasks, taking action, and learning as they go. By the end of this video, you'll understand exactly how they work and why they matter for your business. Let's be honest, your team spends too much time on repetitive work, whether it's categorizing customer messages, scheduling meetings, updating spreadsheets, or following up on leads, these tasks pile up fast. Research shows that 40% of a typical workday is spent on non value added work. For small teams, this is especially painful because the same person handles customer service, scheduling, reporting, and follow ups. You need more capacity, but hiring isn't always practical. That's where autonomous agents come in. Here's the key difference. A chatbot is like an assistant who answers your questions. An autonomous agent is like an employee who completes entire projects. Chatbots respond to prompts. Agents plan multi step workflows, use tools, and execute tasks independently. They reason about what needs to happen, break it into steps, and take action, and they learn from each interaction getting smarter and more efficient over time. Most importantly, they work around the clock, so your business never stops. Let me walk you through exactly how an autonomous agent works. First, you give it a goal like process customer support tickets. The agent understands what you want and what it's allowed to do. Second, it reasons about the best approach and breaks the task into steps. Third, it uses memory to understand context. Maybe it remembers similar tickets from last week. Fourth, it executes by using the tools available to it, sending emails, updating your CRM, pulling data from databases. Finally, it learns from the results and improves next time. This entire cycle happens automatically without you checking in. Now, let's talk about what this means for your bottom line. Businesses that deploy autonomous agents report up to 40% productivity improvements. You're not hiring more people, you're increasing output from the team you have. You're operating costs stay predictable while capacity grows. That's the magic of agents for small businesses. Your team stops doing repetitive work and starts doing work that requires human judgment and creativity. For example, instead of manually following up with every lead agent handles that 247. Instead of your team answering the same customer questions repeatedly, and agent handles first line support, your people focus on closing deals, building relationships, and solving complex problems. Here's your action step. Identify one repetitive task your team does every single day. It could be customer follow ups, data entry, scheduling or reporting. That's your first agent candidate. Start with a pilot project, automate that one task and measure the results. You'll see time savings, cost reduction, and team satisfaction improve immediately. Once you approve ROI on that first agent, you can scale to other workflows. The businesses winning right now are the ones automating their operations today. Your competitive edge isn't about working harder. It's about working smarter. 16. Exploring the Manus Platform: Hey, guys, in this video, we will talk about Manus. Manus is autonomous AI agents created by Chinese company. This is really amazing evolution of all AI tools together. It can autonomously solve a lot of tasks that you need previously create very deep and very detailed instructions and prompts and right now, it can be done totally from scratch. Manus gives you possibility to conduct really amazing deep researches, create websites, make easy operations inside browser, for example, to login form, go and book restaurant or find you a hotel for stay during the vacation. My case studies of using Manus is pretty straightforward. I have task, for example, for deep researchers and it gives me a lot of valuable data during the deep researchers. For example, I need to find tools for Geo just a few minutes of deep research, and I've got detailed report in the PDF, comprehensive comparison of the tool, the description of all main tools and features, et cetera. If you talk about the deep research, you have few free options during the day, you don't need to start paying compared with GPT. So if you need to get really deep research, you need to go to Manus. It can be done pretty easily. You can go and try with some really easy stuff. For example, I've tried Manus for registration on my Ukrainian website. We don't have Capture, so the registration wide my personal data was done and it gave me thank you page in the end of this process. An easy stuff when you need to go to fill the data and finalize the registration, it can be done by and of course, if we need to get more ideas, how we can use manus, we can just go and ask Manus to make a deep research about what are the main key studies of application, for example, of manus in different spheres in business and marketing and we have ideas about onomous task execution, knowledge, and memory retention, complex, workflow automation, API usage, et cetera. A lot of my friends who are more on the developer side, IT side, they use mans for code writing for connecting different services with API or webhooks, creating additional small websites, application, et cetera. Let's see what data we have here on manus with the official website. Here we have the examples. You can just click to explore the use cases and see what we have. For example, if you need to make some scientifical research, you can easily see how Manus is helping with the detailed information and you see the whole process, how it works, how it's doing research, where it's find, et cetera. The final stage, you can go to the a clicking button, skip to the results and you can see the final report how the electromagnetic field is working. You have the detailed PDF with all references, et cetera. As I mentioned previously, some basic website building is really easy to create with Manus. Here we have tangle social guide site and this how it was looking like during the replay. We just ask Mans to research tangles social. Are you master website builder and award winning writer, build a website about how to use tangle social and here we have the basic testimonial. If we are skipping to the result, we have the technical social guide website. We can go and see the whole website if it's needed and we see the O files and if we will upload this on server, we will got the website and here with the link we see how it look like. It's definitely not fantastic, but for website that was created for 5 minutes, definitely, it can be used as a testimonial for the approach. Let's proceed with the example example which is pretty closer to my educational niche. For example, I need to create a course and here we have a testimonial how to create a fast API course we are looking for on replay and we are looking on the basic prompt, I'll create comprehensive fast API course and all requests which we need here. When we are keeping to the results when we already scanned all database where we have this information, et cetera, we can see how a detailed such course can be. For example, here we have the detailed comprehensive faster API course. We can go to the module, start IPI. We have this model, we have additional materials. We can move from one module to another module, so it looks really amazing. Let's proceed such examples can be really we can find a lot of different examples for web applications, for mobile applications, websites, courses, et cetera. If you talk about my thoughts about mans differently, it's next big thing in our regular approach with AI because when you ask Manos to go to register somewhere, to do some monkey manual job for you. I definitely can be game changed in different processes. Manos can create, it can analyze, it can research, it can code. You have standard mode, you have high effort, and this is even in the free version. Definitely, this tool will evolutionize further. It will add additional features, but right now you can see that with manos, you can make really amazing things, really amazing stuff. I will see you in the next video. 17. Getting Started with Manus: Is the manus dashboard. It's designed for simplicity. This is the input box. Here you can operate on the principle of autonomous task execution. You don't need to break down the tasks into tiny steps. You give it a goal and minus plans, executes, and delivers the final result. Below the prompt, you'll see core function buttons. These are the specialized tools that Manus uses to complete your tasks. Let's quickly run through them. This is an image function for generating marketing visuals, social media graphics, or product mockups. This is a slides for creating a full structured presentation, a PPT or a slide deck from a single simple topic or a document. Here is a website for marketers needing quick landing pages, microsites for campaigns, or even full stock web applications to support a product launch. Manus is powered by advanced AI language models enabling you to build AI native application with built in capabilities for chatbots, image generation, and autonomous task execution all without writing code. Finally, you can see spreadsheet and visualization are essential for data driven marketers. Manus can process raw data, clean it, perform complex analysis, and generate interactive charts and reports. Together, these tools allow Manus to handle virtually any knowledge work task in your marketing stack. Thank you for your attention. 18. Manus in Your Daily Workflow: In this video, we will talk again about Manus, but more about practical approach. As you've already understand the capabilities of this AI agent, they are amazing and they can be used in different spheres. I will test it in the most regular stuff I usually use for web development, I will create a basic landing page for our SAS digital marketing agency sales booster. I will show how it helps me this regular task on the leap research and we will do it on practice. Let's start with the basic idea of the web design. Of course, as in any AI agents, it all starts with the sum systematic we need to go to Manus and ask to act as professional website builder. After that, we are going through the website creating process, we need to give the context and we ask to create a website for SAS Boost of digital marketing agency based on uploaded deck. We upload in here at the deck and the brand colors and information from the deck. The last but not least in this story, we need to see how it will deploy and it will be a question when they're asking about how personal this information is, can it deploy and public the website on the internal URL, theoretically, everyone can get it. If you have any personal information, you can say no and stop this process and just get the files. In my case, I was interested to see how it is looking on the final results. I asked manus to proceed. And you see how the process is going on. The first it was extracting content from the brand colors and uploaded PDF. It got the basic information which it is needed. I got the whole information from the PDF, saved images, viewed it, and so on. As a result, it's successfully extracted content and brand colors from the south Booster PDF and initialized reactemplate for the website. And give the brand collars like blue, red, orange, dark blue navy with gradient backgrounds. It's got the company information that says booster Sal digital marketing agency with 1.5 successful projects and basic portfolio, partners, et cetera, all this data was getting here. After that, it started the website creation. You see how the design implementation was going on. After that, it's test websites for responsiveness and visual appeal, validated content accuracy, deploy website and provide access link to user. As I've mentioned previously, there was the request, allow Manus to deploy the website publisicly and here we have yes allowed the text. After all this process was done, we have the public link to the website details we have here, it's on the mano space. We can edit it if it's needed. We can set to private. After we've checked the results, we can ask to other activity and if we need to change it to subdomain, we can move to the prop plan and do. Let's see how our website is looking like now. SAS Bostrom, the colors are arrived. So issues with logo, but it's easily fixed so we can make it. We have design for SAS about our background, we have 1.5 successful projects and since 2008, we have our credentials, Meta Business partner NglePner trusted by 40 plus SAS companies. Why choose us proven reputation, successful portfolio, trusted globally, all these deck information and we have the form and we have the places where I need to push the links at the locos and change something if it's needed. As we see in practical task when we need to do something with this stuff, it's really amazing. As I managed previously, I can manage this website, so I can go here and set it as private. If I don't want to share it, I have all my files that I can push on the server and upload. So examples Wooster website, index CCS, brand colors, South Wooster logo, et cetera, I can take it and publish it on server, so it can be easily done. You see how easy it can be to create some basic lending page with Man. For a few clicks, you got this amazing result. You can try to do it by yourself. Right now we are coming back to another amazing picture of me is the possibility to do really deep researches. Just to recall, you have limited number of deep researches in CGPT. You have pretty good deep researches in free version of grog. But for example, if I need to make a deep research in the mammos, I can ask to just go and find the similar tools with the specifications. When I've named a deep research, you get the whole information with details, how it needs to be done, what and you need, et cetera, so it easily can be done for any activities. So we have here the feature unique selling points for scrunch scrolling down, finding the information, searching for the database. And here we have the analyzed data for all these markups, analyze websites with the similar toolkits, and as a result, we have the list of the competitors. We have the reports inside. We have the detailed, I don't know, 18 pages report with the link to the all competitors description of what is amazing in these tools, why they're good and why they bad. We have the comprehensive comparison. So we can also go through the detailed comparison and find all decks. Here we have even the comparison matrix for understanding what is going on here. For the most important comparators, we have detailed features analysis and seven points. We have it for scrunch, we have for profound, but we have it for all other tools, and if you need, you can switch between these tools, compare them, get more insights, et cetera. The last thing that we can play with you doing this, we can ask it to find a hotel in port for exact dates. For example, let's play it as for the personal life. I can ask I need act as professional tour agent, as professional tour agent. Find a hotel for two adults adults and three kids three, eight and 10-years-old. With minimum rating 8.0 on booking, 1 kilometer nearby city center and budget limitation 200 euros per night, for example. We can give the dates, for example, that dates 8-6 to eight of June 2025. Here we have the hotel research process and it will try to do it. We will see how it's starting and how the AI agent usually works, it needs some time. I'm worried that maybe my credits and mons are finished, it's okay. We have the dataset. It's connecting data source, so it's starting. Is a search process. It has basic search results and here we can find on the booking the details. So it's searching and booking. We can pose it and right now I'm spells hotel in port today, it's an important fact. And it will be implemented better. You see how is the agent browser is working on. It's adding details, it's updating the plan, it's searching the information about porton booking.com. We can see all basic elements here if it's needed. We can see what is searching on Porto on the booking. You see the process, how it's going to booking.com, how it's adding portal, how it's adding check in date, checking out date, how it's adding kits and the age of kids, what information is actual or not, et cetera. All in real time, which is really amazing, you have the criteria that you've already set up. I will make sorting for you, for example, with the filter 1 kilometer close to the city center and it's grabbing the dates. It's already grabbed the dates for our stay. Further, it gives you all details which are needed. I want to spend your time and definitely highly recommend you to register manus and play it by yourself. You see that during today's practical lesson, we went through the website creation and we've created a website for ourselves Booster agency. We've made pretty good deep research and we found all competitors for one of the geo services with the deep description. We started my practical research for the hotels and right now I will have a quick solution for my booking on the booking co. Can do it and try by yourself. You will get the ideas what hotel will be the best solution. In our case, we will have these ideas and you can try Manus by 19. The End of Manual Work: Welcome to the end of manual work. In this lesson, we're going to explore how automation and AI are fundamentally transforming the way businesses operate. Whether you're a business owner, a project manager or someone managing operations, this is about reclaiming your time and focusing on what really matters strategy, growth, and innovation. By the end of this video, you'll understand the real cost of manual work. See the opportunities automation creates and have a concrete action plan to start automating today. Let's dive in. Let's start with the real problem. According to IBM research, poor data quality from manual data entry cost organizations $3.1 trillion annually. Think about that for a moment, but it's not just about data quality. When you rely on manual processes, you're locked into a linear cost structure. If you need to process twice as much data, you need twice as many people. Your costs scale with volume, but your revenue doesn't. This creates a ceiling on growth. Additionally, manual workflows mean delayed insights. Your team is generating reports weekly or monthly when the business needs real time visibility, and perhaps most importantly, your best people, your strategies, your problem solvers are spending 40 to 60% of their time on repetitive manual tasks. That's burnout waiting to happen, and when humans do repetitive work, errors accumulate. A single data entry mistake can cascade through your entire system, creating compounding problems downstream. Here's the good news. Automation is not a future concept. It's happening right now. Organizations that have implemented automation are seeing dramatic results. Automated data pipelines reduce data preparation time by 80%. That means instead of your team spending days cleaning and preparing data, it's done in hours. Across the board, companies report productivity gains of 30% to 40%. In customer support, automation is already saving the industry $11 billion annually. Finance teams are automating 75% of their functions from invoice processing to reconciliation to reporting. And in back office operations, robotic process automation is eliminating 70 to 90% of manual work. These aren't theoretical numbers. These are real results from real organizations right now. So where is automation making the biggest impact? Let's look at five core areas. First, data management. Instead of manual data entry and spreadsheet updates, modern systems automatically collect, validate and transform data in real time. Your dashboards update automatically. You reports generate themselves. Second, customer support. AIPower chat boards now handle 80% of routine customer inquiries. Password resets, order status checks, basic troubleshooting. Your human support team focuses on complex issues that actually need human judgment. Third, sales and CRM. Automation scores least based on behavior, sends follow up emails at the optimal time and predicts which deals are most likely to close. Fourth, finance and reporting. Real time dashboards replace monthly closed processes. Predictive forecasting replaces guesswork. Automated reconciliation replaces hours of manual matching and fifth operations. Supply chains optimize themselves. Inventory levels adjust automatically. Scheduling happens without manual coordination. These aren't separate initiatives. They are interconnected systems that amplify each other's impact. You don't need to be a programmer to automate. There are four categories of tools available today. First, no code automation platforms. These let you connect apps and create workflows without writing a single line of code. Teams using these tools report saving ten to 40 hours per week. Second, robotic process automation platforms. These are more powerful. They can automate legacy systems and complex workflows. Organizations see 70 to 90% reductions in manual processing time. Third AI agents. These are systems that can reason through multi step problems and execute autonomously. They are particularly powerful for analysis, content generation, and decision support. Fourth, specialized tools, meeting transcription services, content generation tools, document processing systems. Each one solves a specific problem. The key is not to try to automate everything at once. Start with one tool, measure the time saved, then scale to three to five integrated tools that work together. Here's your action plan. Step one, identify. Look at your week. Where are you wasting time, data entry, copy pasting information between systems, following up on emails, generating reports, write down your top three time wasters. Step two, automate. Pick one of those processes. Research a tool that can automate it. Many tools offer free tiers, so there's no upfront investment. Implement it. Measure how much time you save. Step three, scale. Once you've proven the concept with one tool, add two to four more. Create an ecosystem of tools that work together. The payback period is typically three to six months. You'll reclaim ten, 20, sometimes 40 hours per week, reinvest that time into strategy, into growth, into the work that only you can do. Don't try to automate everything at once. Start small, measure results then scale. That's the path forward. The end of manual work isn't coming in the future. It's happening right now. The only thing standing between you and a more efficient more strategic operation is action. Start today. Thank you for watching. 20. Automating Data Collection and Analysis: Welcome. In this lesson, we're going to tackle one of the biggest hidden time wasters in modern business, manual data work. Whether you're a business owner juggling multiple responsibilities or a project manager coordinating teams and tracking progress, chances are you're spending hours every week copying data, fixing errors, and waiting for reports. In the next few minutes, you'll discover exactly what data automation is, why it matters right now, and most importantly, how to get started without needing any technical skills. By the end, you'll have a clear practical action plan to reclaim five to 10 hours every single week. Let's get into it. Let's start with the problem because it's bigger than most people realize. Think about your typical week. How much time do you spend copying data from one system to another, fixing inconsistent or duplicate records, waiting on a colleague to send you a report before you can make a decision. Or worse, making a decision based on data you later found out was outdated. If you like most business owners and project managers, the honest answer is five to 10 hours per week. That's a full working day every single week, lost to tasks that at zero strategic value. Here's what the data tells us. 70% of businesses still rely on manual data processes. One in every 300 data entries contains an error, and those errors cascade. Bad data leads to bad reports and bad reports lead to bad decisions. IBM research estimates that poor data quality cost businesses $3.1 trillion annually in the United States alone, and analysts spend 2% of their time just cleaning data, not actually analyzing it. The good news every single one of these problems is solvable and you don't need to be a data engineer to solve them. So what does data automation actually do? Let me break it down into four steps that happen automatically without you lifting a finger. Step one, collect. An automated system pulls data from all your sources, your CRM, your spreadsheets, your project management tools, your databases, and brings it together in one place. No more manual exports or copy pasting. Step two, clean. The system removes, duplicates, standardizes formats, and fixes inconsistencies. Your data becomes trustworthy. You stop second guessing whether the numbers are right. Step three, combine data from different departments, sales operations, finance, marketing, gets merged into one unified real time view. Everyone is working from the same source of truth. Step four, deliver. Insights are automatically sent to dashboards, reports or directly to your team. You get the information you need when you need it without asking anyone for it. The results are significant. Organizations using automation report a 45% increase in operational efficiency. Automated workflows reduce data errors by up to 80%, and insight generation becomes 30% faster. This isn't incremental improvement. It's a fundamental shift in how your business operates. Let me show you how this plays out in practice with three scenarios you'll probably recognize. First, ecommerce owner. Before automation, they are manually pulling sales data from multiple platforms every week. That's 5 hours gone just gathering information. After setting up an automated pipeline, all sales data is collected daily, cleaned automatically, and the dashboard updates itself. They say 5 hours every week and can spot trends in real time, which means faster pricing decisions and better inventory management. Second, a project manager before automation. Collecting time logs, task completion rates, and resource utilization data manually takes 3 hours every week. After automation, an autonomous system gathers everything from the project management tools and generates weekly reports without any manual input. Bottle necks get flagged automatically before they become serious problems. Third, a marketing manager. Before automation, compiling metrics from email platforms, social media, and ad networks takes 4 hours every week. After automation, an integrated pipeline collects all the data and the system identifies which campaigns are performing and which ones are draining budget. The result, faster optimization and a 25% improvement in return on investment. These are real achievable outcomes, and the return on investment is typically reached within two to three months through time savings alone. Now, the most common question at this point is, do I need to be technical to set this up? The answer is no. You don't need to write a single line of code. There are platforms built specifically for non technical people. Zapier connects over 6,000 applications visually. You simply choose what triggers an action and what happens next. Make offers a visual workflow builder with more complex logic for advanced scenarios. Air table combines the familiarity of a spreadsheet with a database and built in automation, and increasingly, AI powered agents can learn your specific workflows and adapt to them over time. Here's a simple five step path to get started. First, identify one repetitive task that's consuming three or more hours every week. Just one. Second, choose a tool based on the systems you're already using. Zapier works with almost everything. Third, build a simple workflow using a pre built template. This typically takes 15 to 30 minutes. Fourth, test it on a small scale before rolling it out across your whole team. And fifth, measure your results, track time saved and error reduction. Here's an encouraging fact. 60% of automation projects are completed by non technical staff. Average setup time for first workflow is 30 to 60 minutes, and free or low cost plans are available from most platforms. This is accessible. This is doable, and it starts with one decision. So here is your action plan. The steps starting today. One, do this today, audit your week, take 30 minutes and list every repetitive task you do manually. Estimate how much time each one takes, identify your top three time wasters. This simple exercise gives you clarity on exactly where to focus first. Step two, do this week. Pick your first automation. Choose one task, just one that takes three or more hours per week. It should be repetitive and rule based, something that doesn't require judgment calls. Good examples are data entry, report generation, file organization or status updates. Pick one and commit to it. Step three, do this next week, get started. Sign up for a free trial of Zapier or air table. Find a template that matches your workflow. Set a 30 minute timer and build it, then test it, measure the results, and acknowledge the win. Here's what the research shows. 80% of automation projects succeed when you start with one clear high impact task. That first win typically leads to three to five additional automations within two months. One final thought worth remembering. Automation isn't about replacing people. It's about freeing your best people to do strategic work, the work that actually moves the business forward. Your team doesn't want to spend their day copying and pasting data. They want to solve problems, serve customers and grow, start small, measure results, scale what works. You have everything you need to begin. 21. Building Applications Without Code: This practical lesson, we're moving from theory into execution by building a complete professional web application all from a single structured prompt. Our focus today is on a real world marketing need, launching a high converting landing page for a webinar titled Next Gen Content Marketing. Drive conversions with AI and strategy. This is the kind of task marketers face all the time. But instead of handing it off to a developer or struggling with no code tools, we'll use Manus to build the entire system for us. With Manus, the process of building a web app shifts dramatically. You'll no longer need to write code or understand complex development frameworks. You just describe what you want to build clearly and specifically, and Manus handles everything else from front end design to back end functionality and life deployment. Let's walk through how this works in practice. Here's the minus input box, and here's the prompt we're using. Create a professional landing page and registration system for a webinar titled NextGen Content Marketing. Drive conversions with AI and strategy. The page should include detailed event information, a speaker bio, a registration form with email validation, and a secure admin dashboard for managing attendees. The design should be clean, modern, fully responsive, and deployed to a public URL. Once you paste this into minus and submit it, the agent gets to work. It begins by breaking the task into key parts, layout design, form creation, database setup, and deployment. It builds the entire app in a single coordinated workflow. Now let's look at the result. In just a few minutes, we have a fully functional landing page that features a strong headline, detailed information about the webinar, a biosection for the speaker, and a clear value proposition to drive sign ups. The registration form is fully validated and all user data is stored securely in the back end. Behind the scene, the app includes an admin dashboard protected by login where organizers can view all registrant details and export the list for email marketing. It's everything you'd expect from a production ready tool built from one single well defined prompt. You can now use this landing page for your own campaigns. Just plug in your content and share the link. No coding, no manual setup, no jumping between platforms, Manus handles it all. Thank you for joining this lesson. Try adapting the prompt to your own topic and see what kind of next gen content experiences you can create powered by AI and driven by strategy. See you in the next lesson. 22. Automating Your Email Communication: This lesson, let's see how to automate a mail Manus. This feature is designed to eliminate contact switching by turning your email client into a powerful agentic command center. Instead of manually moving information from your inbox to a task manager or research tool, you simply forward send or CC a unique Manus email address to trigger an action. Now we will walk through the three essential steps to connect and start working with Mail Manus. The first step is to establish the unique email addresses that Manus will monitor for instructions. These are your receiving emails. Let's go to mail manus setup area within your Minus account settings. You will be prompted to pick a unique email prefix with manspot. Think of these prefixes as dedicated inboxes for specific types of tasks. For example, you might create research manus spots for complex research and summarization. Task four team manuscript for converting emails into structured to do lists. Summarize manspot for quickly summarizing long documents or threats. By creating multiple prefixes, you can direct specific types of work to Manus, ensuring the agent knows exactly what action to take based on the address you use. To ensure security and proper task attribution, Manus needs to know which email addresses you will be sending instructions from. These are your sending addresses. In the Manus mail setup area, locate the section for linking sending addresses. Enter your email you use for work, personal email, a team alias, et cetera. So I will use an imaginative email here. Manus will likely send a verification email to each address. Follow the instructions in the verification email to confirm ownership. This step is crucial for security. Manus will not only process instructions sent from verified addresses, preventing unauthorized use of your agentic workflows. Once your receiving and sending emails are set up, you are ready to use Manus. This execution is simple. If you just got a task on hand, let Manus handle it. So basically what you do is just write a name of your Mnupot. So for example, task for Team and enter your instructions for converting emails into structure to do lists and save. And you can create a lot of these emails for different tasks. The primary method of operation is to forward, send or CC your email to appropriate Manus address. For quick research, where you receive an email with a question you need an answer to, you CCRsearch Manus boot and Minus delivers a structured summary or report right back into your email thread. For task creation, where a messy email thread contains a task for your team, you move forward for task for Team Manus spot, and Manus converts the thread into a structure to do list for your team. For collaboration, when you want to share an email and assign a follow up action to a teammate, you can CC task for Team Manus Spot and your teammates email, and Manus creates a shared task, allowing you and your teammate to build on top of each other's progress. For automation, where you want a recurring report summarized weekly, you can set up an email rule to forward the report to summarize Man Spot, and Manus turns this recurring task into an agentic workflow, delivering the summary without manual intervention. Mail Manus is about achieving a state of from email to done by following these three simple steps, creating your receiving addresses, linking your sending addresses, and using the appropriate action forward, send or CC. You can offload research, structure to do, and automate recurring tasks directly from your inbox. Thank you for joining in this lesson, and I'll see you in the next one. 23. AI Meets Spreadsheets: This practical lesson, we're moving from theory into execution by building a complete professional data management system, all from a single structured prompt. Our focus today is on a common but time consuming task, integrating multiple marketing data sources into a single spreadsheet system that automatically collects, organizes, analyzes and visualizes performance data. Instead of manually copying figures, building formulas or formatting charts every week, we'll use Manus to create a fully automated workflow. With Manus, spreadsheet integration becomes a strategic design process rather than a repetitive manual one. You don't need to create endless tabs or write complex formulas. You simply describe what data you want to track, how you want it structured, and how often you want it updated and Manus builds the system for you. From live data feeds and custom metrics to auto generated dashboards and reports, the agent handles it all. Let's walk through how this works in practice. Here's the manus input box, and here's the prompt we're using. Create a comprehensive spreadsheet based data management system that automatically collects marketing performance data from email, social media, paid ads, and content platforms. Include formulas to calculate ROI, track engagement trends, and compare results across channels. Generate weekly reports as PDFs with charts, insights, and a sharable dashboard. Once we submit the prompt, Manus gets to work, setting up data connections, organizing the spreadsheet, building analysis logic, generating visualization, and scheduling report delivery. It breaks down the requests into coordinated tasks and assembles the system from start to finish. Now let's look at the result. In just a few minutes, we have a fully integrated data system spreadsheet file. Each tab is dedicated to different channels, email, social media, paid ads, and content with live data pulled in through secure APIs. A Master dashboard summarizes all key metrics, spend, conversions, ROI, and traffic, and uses color coding to highlight which campaigns are outperforming, meeting or falling short of their targets. The system also includes custom formulas that calculate growth rates, compare current performance against weekly or monthly goals and flag anomalies. Charts visualize trends over time, such as email, open rate growth, ad click through rates or block traffic surges. You can now manage your campaign performance from a single intelligence spreadsheet system. No exporting, no formulas, and no guesswork. Just clear data updated daily with visual summaries and automated reporting built in. Thank you for joining this lesson. Try adopting the prompt to suit your own data sources, metrics, or reporting schedule. Whether you're managing marketing, sales or operations data, Minus turns your spreadsheet into a smart connected system powered by automation and driven by strategic results. 24. From Tasks to Systems: Welcome to from task to systems. If you've been automating individual tasks, sending emails, updating spreadsheets, triggering notifications, you've taken the first step. But here's the truth. Individual tasks are just the beginning. The real power comes when you connect those tasks into a complete orchestrated system. In this lesson, you'll discover why systems thinking transforms how you scale your business and how to shift from managing isolated automation to building workflows that work together seamlessly. Let's dive in. Let's start with a common scenario. You've automated your email sequences. Great. You've synced your contact form to your CRM. Excellent. You've set up a notification when a deal closes. Perfect. But here's the problem. These automations are islands. They don't talk to each other. Your email system doesn't know what your CRM knows. You notification doesn't trigger the next step in your workflow. And when your business grows, you're not scaling your automation. You're multiplying your chaos. Individual task automation hits a ceiling. It saves time on specific tasks, but it doesn't solve the deeper challenge. How do you make your entire operation run on autopilot? That's where systems thinking comes in. Now imagine a different scenario. A lad fills out your form. Automatically, the information flows into your CRM. The system checks their profile and triggers a personalized welcome email. That email includes a link to a resource tailored to their industry. When they click the link, the system logs their engagement. If engagement is high, and notification alerts your sales team to reach out. If it's low, the system automatically sends a nurture sequence. This is orchestration. It's not just automation, it's choreography. Each task knows its role, knows when to act, and knows how its actions affect the rest of the system. This is what we mean by moving from task to systems. Systems work because of three fundamental principles. First, interconnectedness. When you connect your task, a change in one place affects the entire system. If your email automation triggers faster, your CRM gets updated sooner, your sales team responds quicker, and your conversion rate improves. One change, multiple ripple effects. Second, feedback loops. Your system doesn't just execute task, it learns. If a customer open your email, the system notes that and adjust the next message. If a deal closes faster than expected, the system learns that pattern and applies it to similar prospects. Your automation gets smarter over time. Third, emergence. This is the magic. When you connect individual tasks into a system, something unexpected happens. You get results that none of those individual tasks could produce alone. Your conversion rate doesn't just improve it skyrockets. Your team doesn't just save time, they gain strategic capacity. That's emergence. These three principles are why systems scale better than individual tasks. Systems don't appear overnight. They evolve as your business grows. In the MVP stage, you're focused on quick wins. You automate your email sequences. You sync your contact form to your CRM. You set up basic notifications. These are individual tasks, but they save you time and deliver immediate ROI. This is where most businesses start. As you move into the growth stage, you start connecting these tasks. Your email automation triggers CRM updates which trigger project creation, which trigger team notifications. You are no longer managing isolated automations, you're building workflows, consistency improves, efficiency increases. Finally, in the enterprise stage, you have full orchestration. Your system makes decisions, its scores leads based on behavior. It routes them to the right team member. It triggers compliance checks. It learns from outcomes and adapts. This is a true system, not just a collection of automated tasks. The key insight, you don't need to start at enterprise level. Start where you are, but build with systems thinking in mind. Here's how to start your transition from task to systems. First, audit your current automations, write down every automated task you're currently using email sequences, CRM scenes, notifications, everything. This is your baseline. Second, identify the connections, look at your list and ask. Tasks talk to each other. Which automations would be more powerful if they were connected? Third, map the dependencies in what order should things happen? What must be completed before the next step can begin? Fourth, choose a scalable platform. Don't pick tools that only handle individual tasks. Look for platforms that support orchestration, tools that let you connect task and manage dependencies. Fifth, start small. Pick one workflow. Connect just two or three tasks into a simple system, get it working, then expand. Finally, measure and optimize, track what works, adjust what doesn't. Your system will evolve as you learn. The shift from task to systems is a journey, not a destination. Start today and watch your automation and your business scale. 25. Project Based Automation: This video, we will talk about how to use one of the most prominent features inside manus manus projects. Definitely, if we talk about the functionality of manus, it was already amazing the full autonomous CI agent when it was not even mainstream for everyone who have any code experience. All was good, but the feature that they created just in December 2025 and it was not enough to automate the task without it. It's something similar to basic AI assistance like M GPT, in GPT or Jams inside Gemini or projects inside, Cloudenthropic. Right now we have the projects as one of the crucial features that help you automate the tasks, automate the regular proms you have and make the possibility to implement it on the regular basis. So if we talk about the Manus project, it's definitely is amazing feature. It's something like saved prompt. You already have the prompt that you like. You leveraged it and right now everything you need to do, you need to save this prompt, add additional materials, add a knowledge base, and you will get what you want. Let's see how to create the project. You have all projects in the left menu. You can make it like this closed or open. You can just click create a project and here you have the basic data. You can add the project name and the project instructions. The same way as we are doing it in any of the basic system. I recommend to structure a prompt with Rs. What do you need to do? What do you need to analyze? All this stuff is described in instructions. After you will save the project, you also can do additional fine tuning. You can add some additional materials, conduct your knowledge base, et cetera, all the stuff it definitely can be updated. Let's go and see what type of projects I'm using right now and what do we need to do to make them really perfect. First of all, one of my favorite projects it's my video presentation creator, here we have something pretty straightforward and similar. We create a lot of theoretical lessons and to create something theoretical, you need to add, I don't know, some idea with research or support article and after that, transform it in video. How to leverage this stuff, it's pretty simple. First of all, you need to give the detailed instructions. In our case, we ask to go through the next flow, first of all, research, research one based on the data we gave, for example, PDF with some research or link to the Help Center, et cetera. It transforms research in slides. Manus use pretty powerful model nana but nana, maybe one of the best presentation creation models right now. It transforms Isn slides. Slides transformed in script. We have the script script pushed to audio. We use here the MCP of 11 labs and we have the video. Here we have the raise prompt, which describes everything, how it works. We have role and all details action, the final description of how the examples look like all in one place. Here how it works. If we talk about utilization, definitely what we can add here, we can add additional files. All videos and all presentation looks in the same style we need, and if we talk about one of the most prominent features here, we have the connector we have element labs to voice over these presentations with my personal voice and this how it works. After that, we see how the research was conducted, how the slides were created. The sides were structured and transformed inside menus. After that, they were created inside nana banana and we see each slide and on this stage, we can post production, double check, and add any additional features if it's needed. It goes to allowing labs through the connector and create the voice at the final result, we have synchronized video. I have course Builder which optomates a lot of workflows. I have my digital marketing strategy helper maybe it's one of my most favorite where we grab all information which we need for the competitors analysis. For example, I want to analyze any big website. For example, I can show it on the example offby.com. I give just website link and after that, it goes and conduct the whole research. For example, in this case, I will show you how it works on already conducted research. Here it is. So it found for me the company mission statement, Company USB, I done competitor analysis based on the similar web data. It gave me some smart goals, target audience portraits, pretty digitalized content plan. Example of the post images, captions and the budget in the media plan for this stuff. If you need any references, it will give you also some references, so you can go and work it here. What is amazing inside Manus after you used it? You can go to the settings and in schedule task, you can add and schedule any task based on these projects. It can be just regular chat. It can be exact projects. So for example, if you want to scale, you have the title, you have the prompt, and in this case, based on all connectors, you can always get the best result on the regular basis. For example, I've created a special project that monitored regular tech crunch. We are interested to get the most freshest information about any major AI updates, one model add Samson, et cetera, and I gave me possibility to have evening digest every time when something is crucial happening inside AI world, I don't need to go to crunch or any other website personally, I need just to ask a scheduled task and I will get it on my personal email. 26. Building Multi Agent Workflows: Hey, dear. In this video, we'll talk about practical usage of agent mode in HGPT by Open EI. Definitely, it's the most simple example that everyone loved to use in browser AI agents, but we will show how to use the tool I commend. What are the pitfalls, what can be pros and cons, et cetera. After that, we will go in more deep business marketing and development examples. But this is just basic showcase how to. I just took the prom that we had in our presentation, plan my three day trip to Tokyo from Lisbon for next month with $2,000 budget. I preferred cultural experience and good food. That's the example that we had in the previous deck. Everything we need to do, we need just to go to the agent mode. Here you see how many usage you have for these months and it's always important to remember that this is limited feature for plus users. You still have it in pro and teams and enterprise, but it's not accessible in free plan at least so we are turning on agent mode and we can start. We are going forward and see how it's going on. So it's setting the desktop at the first stage where you see how it's going on. If you need, you can click and you will see the reasoning. You see that yes, it's asking about budgets. Okay. Let's take flexible dates and it will give you more flexible. If you will see that something is going wrong, you don't have enough data, et cetera, you can easily go through. We see the Lisbon to Tokyo round trip, August 2025 flight costs. And it right now is going and it's right to find all the information. You can take over the browser, if it's needed, you can stop. You can go through the activity and see how it's working in the process. All steps are required here. We can come back to the desk review and we see the flights from Lisbon to Tokyo. It's searching Google flights right now, going through the reading mode, it's trying to find the price. Definitely with such small limited budget as two K, I will need to find very cheap flight and it's an issue because the average flight is 800, one k euro. We see that it's really hard to find cheap flights in such limited time when you are planning something in the month. But trying is going to Expedia, is going to Japanese airlines, it's searching how to find such places in red, et cetera. And it's going on how to plan for days in Tokyo, Japan, it's still researching. You see how it's doing the research. It's recommends me to visit Tokyo National Museum with private tour. Maybe after that it will give me some more details. That's still working, you see the whole process. After that, it's going to read it to find the best recommendation here. Red trick advisor YouTube, all sources are working on. Honestly, I don't like sushi. My kids like sushi, but Tokyo's sushi is a good idea. To visit Toki and don't taste sushi, I don't think that it's a good idea. Amen, amen is also a very famous Japanese soup. Let's go through ramen experience outdoor market. You see all stages. If you need the documentation, you can go to the activity to double check the link, so you can go back to each link to see how it's going. If you need to come back to the browser and check this article, you can take over the browser, you can stop searching if it's needed. But the basic idea when you have some such plan, you can just give the task to hGPT open mode, and close your laptop, go for a walk, wait a little bit, and come back to it after the job will be done. So it's still searching hotels. With my budget, I don't have a chance for four star of five stars, but three star maybe we'll find something and it's searching budget year tour. It's still searching the solution, maybe we will have something and it's going to find Tokyo subway details, public transport. All recommendations for me, we have go Tokyo ticket page. Maybe not the worst solution, so we have a possibility to double check this information. We are looking information here. Explaining our trip for further steps from Naritas to Shinjuku, the best transport options in Tokyo. Usually, it takes 5-15 minutes, depends on the capacity, on the tasks, et cetera. You definitely have the basic understanding what do you need to do here. You see what do you need to change, what experience you will get when we come back to our prompt, we understand that it's pretty fluffy basic prompt and the more context you will give here, the better it will work. If you even restructure this prompt with race, our usual basic framework with the role, action, context, execution is definitely will be better. But even with this basic prompt, it makes some assumptions, it's searching for it's even found 465 box flight which look like impossible, but somewhere it found it. We will double check these flights after that. This looks like more regular flights. This pricing is more similar to this. In our further approaches, we definitely will go and work with more details. But right now, you see the basic principle that you have your autonomous browser agent, which is searching something, finding, researching, making solutions based on it, giving you link. If you will say, yes, go here and book if it want to be capture or two factor authorization, it will do it for you, you have the whole magic of EI in your hands. Yeah, it's Tokyo Museum. Honestly, my kids were asking about Japan, maybe not this year but next year, it's time to make their dream come true. It's found information about Tokyo National Museum. It's double checked the opening hours. It's restructured everything. I found the date on the trip article. And went forward. It's searching for the tickets. All the data will be done in this list, so you will have all in one. Definitely, it's pretty useful for you to go through. You'll see the whole process here. It's searching the data where it's needed. As I've mentioned previously, right now, it's doing calculations. It's doing calculations, it's putting all the details together. Like in deep research or 03 model, you see how it's structure, what calculations were done, what metrics it uses, et cetera. It's really amazing at this stage. It's creating the detailed overlook. You see all elements of this trip, museum ticket, food cost, total activities cost, flights. I think that just in a few minutes we will get the final file with this information. If you need a deck to show it to your family or friends or team when you are planning on something, you can go forward and act it. Here we are here we have our three day Tokyo trip plan. It took us just 10 minutes, not so long. As I mentioned, in average, it's 5-15, but for some complex tasks, it really can go half an hour. We see the flights, flights definitely the most expensive part in this story, travel time, airport transfer, hotel, I gave us hotel details, public transport, local IC card, daily itinerary, with all details with food, with historic ways, where are we going, how are we going? What are the approximate budget summary per person and tips and cultural etiquette. All details are here and you can request details if it's needed. As I mentioned, it's the basic way how you can use agent mode. We didn't ask it book something, but it can if it's not captured there or if you don't have some Um limitations with authorization, two factors rezation et cetera. But my recommendation here is to try to play to see how it works for you. In our next videos, we will go to more business use cases. Feel free to ask questions at this stage, and we will add more details to this course after agent mode will evolutionize, and open EI will add here more details and more results. So see you soon and don't forget not just watch these videos but practice, practice, and practice, and get great results with EI. 27. Course Summary: Welcome to the course summary for Building basic AI agents. Over the time you have spent in this course, you have gone through a complete transformation from someone who uses AI tools to someone who understands, builds and deploys them. Let's take a moment to map out exactly what you now know and what you are now capable of. Five core skills, real practical and ready to use today. The first skill you build is understanding what AI agents actually are and why they matter. You learned the fundamental difference between a chat board that simply responds to your questions and an AI agent that takes action on your behalf, agent plans. It executes. It works through multiple steps independently without you having to guide every move. You also explored the modern AI landscape, GPT four from Open AI, clawed from anthropic Gemini from Google. You now know what each one does and why 2026 is a turning point for anyone who works with information. The key shift you made from talking to AI to AI working for you. The second skill is communicating with AI effectively, and this one changes everything. You discovered that the quality of what AI produces is almost entirely determined by how you ask. Great prompts are built on three things clarity, context and constraints. Specific about what you need. Give AI the context it requires, who you are, what the goal is, what format you want, set constraints, define the length, the tone, the audience, and what to avoid. And when the first result is not quite right, iterate. Treat it like a conversation, not a one shot command. This skill alone will make every other AI tool you use dramatically more powerful. The third skill is building custom AI assistant, tools you designed yourself for specific jobs. You learned how to identify the task in your work that are perfect candidates for AI delegation. Then you built four real assistants, a video content optimizer that generates professional prompts in seconds, an image creation assistant that speaks the language of AIR generators, a customer support agent that handles inquiries automatically around the clock and an appointment scheduler that manages booking requests without any human involvement. These are not demos. These are working tools you can deploy right now. The fourth skill is knowing which platform to use for which job. Custom GPTs are fast to build and great for writing and content tasks. Cloud projects excel at research and analysis with deep nuanced reasoning. Google Gems integrate seamlessly with Google Workspace, and autonomous agents handle complex multi step tasks entirely on their own. You now have a decision framework, simple task, custom GPT, need analysis, Cloud complex workflow, autonomous agent, the right tool for the right task. That is the skill you have built. The fifth and final skill is the most powerful thinking in systems, not just tasks. You can now automate data collection and analysis. Build web applications without writing a single line of code, automate email and communication flows, design multi agent workflows, and orchestrate entire systems that run themselves. So here is your challenge. Pick one workflow, something you do manually today. Automate it this week, start small, prove the value to yourself, then scale. You have completed this course with a genuine practical skill set that most professionals do not yet have. Now an AI powered professional. Use it.