Transcription
You, guys, everyone hello. Hello, hello. How are you doing? How is your mood on this summer Wednesday? Ah, so I noticed the poll. How often do you use it in your work, guys? Well, about a couple of times a week, 50% answered. It seems a bit too little to me. And I, for example, can no longer imagine my, uh, activity without artificial intelligence, because I solve many tasks, most tasks, precisely with the help of artificial intelligence at the moment. And, of course, ah, yes, it's just a part of my professional activity. But how is your mood? And I want to ask additionally to the poll that was. Why? What is, probably, at the moment, ah, Pasha, by the way, people are writing comments, when did it not start on YouTube? M, no, everything is going. Everything is going. Perhaps about the start. I don't know, about what start. Clarify, guys, what start do you mean? Yes, I want to ask, perhaps, in addition to the poll, what is stopping you at the moment from fully implementing it into your processes. We will also talk about this partially today. Ah, but I would also like to hear from you. Here. Ah, the topic of today's broadcast, first of all, uh, probably, the most important thing, yes, is to create such an AI assistant, which will be, uh, a kind of eavesdropper in chats with clients, because for me, for example, this is a big problem, and honestly, it depresses me a lot when I have to spend another, uh, 20, 30, 40 minutes, uh, to collect everything from the chat, structure it, turn it into some coherent backlog. This is one point, yes, that triggers me. The second point is that sometimes there are problems with clients, yes, meaning you agree on one thing, then the client forgets about it, and the agreements were not fixed, and so on, and you have to search through correspondence, respectively, the client's message, what he said there, and so on. forward it to him, literally shove his own words in his face. And yes, this also eats up, probably, not so much time, but it burns out, uh, your ass, if I can put it that way. Ah, therefore, yes, our main task today is to create such a small AI assistant, which can be added to a chat with a client. This is a very convenient thing. Ah, it will listen to every message that you write, that the client writes, that some other people write, who are in this group. Ah, all this is collected in a database. Ah, and then from the database, we can get a complete summary, pack it into a backlog, or even into a commercial proposal, if it's some new work and so on. That is, there are actually many possibilities. And you can improve this agent infinitely. Ah, Pasha writes, "I already want it, now 40% are not up to date. We haven't sold anything yet, but they already want to buy." I understood. Okay. Ah, so Tatyana Subotina writes: "If the client doesn't write, but only sends voice messages, will the system work?" Yes, ah, we will implement a thing that will transcribe voice messages. So, as I said, besides voice and text messages, you can also develop this agent and add some add-ons for document recognition, recognition of, I don't know, links that he sends, so that this is also somehow noted in the summary that will be collected. So the potential is quite large. How you finish and refine it is up to each of you. Here. Well, and what? Ah, ah, let's start gradually. I want to start not with practice, yes, the whole broadcast will take about an hour, maybe an hour and 20 minutes, but I will try to tell and show everything as quickly as possible. Ah, before we start, I would like to tell you why, ah, we created this project, you could say, the Lazy Method channel. Ah, there is such a, probably, unambiguous answer in connection with progress, in connection with the development of technologies, in connection with the change in certain action patterns. And here I have prepared such a thing. So, to stage, ah, yes, about what I mean by action patterns and so on. Ah, here on the screen we see now, yes, so basically the evolution of how people wrote books, yes, so initially it was all written with a pen, on parchment, on some paper, and so on. Here. Ah, then in the 19th century, we had, uh, typewriters that simplified writing some texts, preparing documents, writing books, and so on. And in the 20th century, we got laptops, which we can take everywhere with us, throw them in a bag and go, sit anywhere and write a book. Well, this is conditional. What do I mean by this? That the action and the result are absolutely the same. Everywhere, ah, everywhere, respectively, ah, this leads to the same result, yes? This is what? This is a book. But the process of creating this book with our technological development has changed. Accordingly, the action pattern for implementing the task of writing a book has also changed. And now, we are, in fact, seeing the same thing. Ah, let's reveal a little more now. Let's start with a bit of statistics. The source, if anything, is Open AI. I just went and asked how, in general, ah, people solve tasks now and how much routine or something else depresses them. To which ChatGPT answered that 67% of professionals spend 21.5 hours a week on repetitive tasks, yes? That is, there are a number of tasks that, for example, Yuris does what? He, uh, prepares lawsuits, yes, a lawsuit is essentially the same scheme, yes, meaning the document architecture is the same, only the name, surname, and reason for the lawsuit change, yes? That is, preparing some documents. Essentially, routine, routine. And it takes over 20 hours for such routine. Yes, here, I think, even more, probably, in practice. Ah, some professionals, an average designer spends more than 12 hours a week searching for references, creating mood boards, and entrepreneurs, businessmen spend 15 hours a week analyzing data and reports. Ah, why, why does this happen? Because work patterns have not been updated for a long time, yes? Meaning, we have some task and we have a certain algorithm of actions on how we implemented it, uh, earlier, yes, before 2022, probably, because the first version of OpenAI, yes, the company OpenAI appeared in 2015, and ChatGPT, the first ones were released in 2022, and after that a small revolution began, a revolution, including in work patterns, yes? Meaning, the same thing we saw at the beginning, yes? The result is the same, but the approach and the algorithm of actions, of course, are changing. And now AI is fully changing the rules and sequence of actions, yes? Meaning, for the same task. Here we spend 3-4 hours. Here we spend 45 minutes. Why? Because we receive a task, we use the right tool, the right prompts in artificial intelligence, we get some structure, we supplement it with expertise, which, by the way, is very important. Ah, and accordingly, the task is ready. Ah, what is most important in all this? The most important thing is, of course, expertise, because, in my opinion, AI, precisely, it enhances your strong qualities, ah, yes, your expertise and so on. We will dwell on this a little more in the next point. But the point is that a linear professional with simple work will, um, probably gradually be reduced, yes? Meaning, undoubtedly, people will lose some work, and this is a minus of evolution, yes, but professionals and experts, like us, in design, in development, in product, in product history, and so on, meaning, thanks to artificial intelligence, we will strengthen our qualities and will, let's say, remain in our niche, if I can put it that way. Ah, what does this mean? That don't be afraid that artificial intelligence will replace you. Why? Because if you have some expertise in some classical area, yes, be it development, design, marketing, SMM, I don't know, sales, analytics, and so on, if you connect AI to your work, you will strengthen all your competencies, you will become faster, smarter, and stronger, and you will remain a valid professional who is in demand in the market. Ah, linear personnel, unfortunately, those who do not have any specific expertise, will gradually, little by little, be reduced, and people will indeed lose their jobs. Therefore, AI is not a panacea. AI is a good add-on to your core competence. And here you can give a good example, it's iron and carbon, yes? Meaning, iron itself, as we see, ah, it rusts, it is quite, it is a soft material, ah, but it has some good, good basic qualities that can be enhanced with carbon. Yes. Meaning, such a chemical bond gives us, ah, respectively, steel, ah, and from steel, we can already build metal structures, bridges, buildings, in short, everything, everything, everything, because steel is strong, there is stainless steel, and so on. Ah, many variations. It all depends on what percentage of carbon is added to the iron. We get a certain steel alloy, yes? Meaning, you can also get cast iron, but cast iron is also, it's heavy, ah, respectively, it has its own disadvantages. But the most, the most fiery, so to speak, ah, in all this is, of course, steel. And artificial intelligence can also, let's take some expert, yes, in some field or, well, in some direction. We add artificial intelligence to it, and we get a kind of supercomputer, ah, a super-brain, uh, which knows everything about everything and can always solve various problems. Ah, I will say right away that in my field, in my work, artificial intelligence has given me a lot, ah, and has greatly strengthened my basic understanding of development, basic understanding of analytics, and some product stories, yes? Meaning, before, I couldn't afford to come up with an idea, go and implement it, because I needed a backend developer, a frontend developer. Now I have Laravel, and I can assemble an MVP in literally one evening, yes? Meaning, an MVP of any product, ah, which can already be tested, given to people to try, and so on, I can assemble it in literally one evening, yes? Meaning, I had initial development, a basic understanding and idea of development, yes? But I didn't have any super-deep knowledge in frontend, backend, and DevOps. Ah, and accordingly, this did not allow me to create products, various projects, and my own ideas. As soon as I added artificial intelligence to my understanding of development, ah, I received many benefits in terms of creating my own products. Ah, including in the channel, I talk about the fact that we are now designing a full-fledged startup, you could say. There is already an MVP, there are already first closed users, ah, who fully use the tools for analyzing viral content. Could I have done this without artificial intelligence? Probably not. Yes, not probably. I definitely couldn't have done it. Ah, but now, yes, now I can implement such projects. Everything is cool, probably interesting. Ah, what do you think about the whole AI thing at the moment? Is it some kind of hype or is it a real thing that can, like carbon and iron, enhance your expertise, your specialty, and everything else? Yes, guys, let's probably wait for your comments. There are people in general, but no one is active, no one is writing anything, so we are waiting for your comments. Support us and answer our questions, please. Yes, it will be great. The more you answer, the more interesting the broadcasts will be. And, of course, we will better understand who our audience is, and accordingly, we will prepare only useful materials for you. Here. Therefore, ah, your activity on the broadcast is a very, very important thing for us. Ah, I will continue for now. So, all this is cool, all this is interesting, cool. AI, yes, from all the irons, everyone is chattering now, everyone is talking about how it will change the world, or conquer it, or fire thousands, thousands of employees, and so on. But I suggest you still adhere to the concept that AI enhances your expertise. But why are many people now not starting to do this? Here again, there are four main problems that experts, professionals, various, um, so to speak, uh, guys from different professional spheres most often encounter. First, I don't know where to start. There is a lot of information on the internet, a bunch of Telegram channels. Today, for example, three news came out that there are 10 new tools that will allow you to do this, that, that. Therefore, yes, here, probably, the main problem is, you don't know where to start, because there is information overload. What to believe, what not to believe, what can, what cannot. Uh, here, respectively, there are very, very, very many questions. Ah, the second point that I have heard among colleagues in general, yes, and in the digital sphere, is that there is no time to study, yes? Meaning, a small paradox, there is no time to save time, because artificial intelligence, ah, this is, yes, such a thing that needs to be studied a little more. Here, Anton writes: "I don't know where to start, it's me." Believe me, when I entered this field, it was also difficult at first, but there are certain frameworks for where to start. We will also talk about this a little today, by the way, yes. Next, ah, what should I automate, yes? Meaning, the lack of basic knowledge and the lack of understanding of what AI tools exist, what can be done with them? Ah, is there anyone, or rather, in the information field, who can tell and show how they automated their activities, for example, implemented artificial intelligence, and what were the results? Ah, too much focus on technology, not on the result. Ah, you know, yes, there is a saying: "The eyes are afraid, but the hands do." Here. And it fits well here. Why? Because we think, damn, this is all super complicated. This is some kind of rocket science. To understand this, you need to be a genius, a developer, to automate something for yourself. You need to know so much and so on. But no one focuses on the fact that by learning something partially, gaining new knowledge, you will save 10, 20, 30, 40 hours for yourself in your professional activity. And a person continues to get stuck in all this, in some routine, in operational tasks, which, first of all, consumes him. Secondly, it starts to demoralize him, because there are tasks that, well, you really think: "Why the hell am I doing this?" I could have gone and made a couple more cases on Behance. Eg, uh, there, post something on Made in Webflow. Ah, I don't know, simply build a portfolio website for myself, but I'm doing this thing, yes, a specific task, which, yes, besides eating up your time, also demoralizes you within your activity, your development, and so on. Ah, and all this is indeed true. And precisely for this, we created this project, method, ah, concept, yes, the lazy method. Why is it called that? Because, um, I believe that laziness is not, it's not like some vice, but it's something that will allow us to not do those actions, thanks to the new tools that have appeared with the development of artificial intelligence, and make us, uh, not so much progressive, Ah, damn, I forgot the word. Not progressive, but effective. Here. Here. Accordingly, yes, that's why it's called lazy method. And through this project, I want to convey, first of all, all those ideas that I voiced today, yes, because I sincerely believe in it. Ah, that artificial intelligence will indeed greatly strengthen us, professionals in some field, experts, and give us the opportunity not to waste time on the unnecessary and to spend time on the necessary, however stupid it may sound, yes? Meaning, when I can free up, even if it's 5 hours a week, not to deal with document flow, not to deal with, ah, some managerial tasks, yes? Meaning, what we will do today is to collect correspondence and structure it into a summary. I can dedicate this free time to developing my own projects. Ah, yes, spend time at school, come up with something new, ah, I don't know, at the studio, and so on, yes? Meaning, I free up my time for creativity, for development, and for everything that will bring me concrete benefits. Managerial work, in context, in perspective, is unlikely to bring me anything. Yes, the client will be super-duper satisfied, but if it can be done cheaper, easier, cheaper in terms of time costs, ah, faster, and the client will be even more satisfied, because you also got back to him faster with a backlog or a commercial proposal. But I believe that you need to grab onto such tools. Here. Therefore, yes, besides the fact that we now have a Telegram channel, in which I try to provide as much content as possible, talk about what I, what topics I reflect on, share with you. I see that there are active commentators. Guys, thank you very much for, ah, you are showing activity. This means we are not here in vain, ah, and we are not preparing all these materials for you in vain. Here, probably, the lyrical digression is over, yes? Ah, we have figured out why we are all gathering here, ah, and, accordingly, why we created this project. Now let's get to our task for today. Yes, why I chose this particular task of organizing dialogues in Telegram, because it is the main devourer of my time. Meaning, look how, first of all, I have several directions in which we are developing, yes? Meaning, there is a studio, there is a school, now the method has been added, now there is a project, uh, that I am doing with the help of, ah, Laravel. There is also one idea about expanding the product matrix in Fluid, yes, and opening the DeepCS division. I want to separate it, ah, the direction of automation, integration of artificial intelligence as a service. I decided to separate it a little from Fluid and launch it as a separate brand. There are many tasks, but at the same time there are current clients, there are current current clients, with whom I have to spend a lot of time. And the task itself, that, uh, a day has just passed, yes, meaning we discussed something, came up with something new, some, I don't know, edits, no edits from the client, all this needs to be fixed, it needs to be, damn, time to sit down, structure everything, form a backlog, or form some new offer, so that, ah, then we can send it to the client for approval, fix these agreements with blood, so to speak, and, accordingly, move on. Meaning, we can write a lot of messages during the day, ah, and the dialogue should end with some summary, yes, and this summary needs to be collected. It is collected quite a bit. You just go through the correspondence, in fact, and that's it. Ah, it eats up time, yes, it demoralizes, yes, strongly, because I could spend these 30 minutes on the next meeting, or dedicate time to my pet project, or dedicate time to forming the product matrix in Deepcons, and so on. Meaning, there is something to do, and I don't want to spend time on this. Therefore, for myself, I decided, from the point of view that this is my main time devourer at the moment, document flow and dialogues in Telegram. Ah, there are ready-made solutions, for example, for Zoom and for Google Meet. This is TLDV - also a bot that records the entire meeting and then sends you the dialogue, can make a summary and so on. I haven't found a ready-made solution for Telegram, so we'll do it ourselves. Here. Ah, but there is also an important point here. In fact, you don't just have to identify your time devourers with a burning butt. In fact, there is a full framework, yes, that can confirm your hypothesis that a particular task, ah, consumes your time. It's called RATP analysis. It stands for Analyzable, Time-consuming, Predictable, and Painful. Ah, and yes, let's break down each of these. Ah, Routine means how often the task of summarizing dialogues and meetings repeats in the context of my task, it repeats every day. Analyzable. Are there clear rules? Yes. Meaning, are there clear rules on how you solve this task? I open Telegram, analyze all, respectively, incoming text and understand that, yes, this task can be solved according to clear rules. Ah, Time-consuming takes 20.5 minutes. Well, here it's clear, it, yes, can sometimes take more. Predictable result is predictable, yes, predictable, because in the end we must create either a backlog, or a summary, or, well, some outcome of the entire dialogue. Ah, Painful, does it irritate or tire, both in my case, it both irritates and tires. Here. Therefore, yes, there are frameworks that allow, in fact, RATP, ah, this is not the only framework that will allow you to find your time devourers, but it seemed to me that it is the most understandable, yes? Meaning, for all questions, we can, ah, quite simply and quickly check, more precisely answer. Ah, what I want you to take away? First, this thing, yes, screenshot all of this now, ah, and after the broadcast or on the weekend, I understand everyone is busy, everyone has current projects and so on, ah, on the weekend, dedicate about an hour, look at your calendar or task manager, look at what tasks are there, and first of all, simply identify which tasks are like weekly or daily, yes? Meaning, are there repetitive tasks? There definitely are. Next, we take this task and start analyzing it, ah, precisely as I have analyzed it on the example of dialogues in Telegram. Analyze it, and you will take the first step towards finding your time devourers and understanding what, ah, what should I do with this potentially, yes? Ah, here, that's all. We can move on to implementation. The implementation is quite simple and fast. I can probably start with the tools. Ah, the tools that we will need are Telegram. Telegram is N8N, of course, the heart of all automation. There is also Make Zapier, but yes, this is not the topic of our broadcast today. Ah, Telegram, Airtable as a database. OpenAI, of course. Well, let's, yes, let's write by models right away. GPT4O. For transcription, uh, for transcribing voice messages, it's Whisper, ah, it is part of the OpenAI infrastructure, so yes, it's just a model that solves this task. Ah, what else? What else? What else? Probably, probably everything, yes? Meaning, these are the tools that we will need today to solve our task. Ah, so, ah, Airtable today, Dn, you asked about why I don't use Airtable, because Airtable is probably the most understandable thing for everyone. Ah, you can self-host everything directly. You can deploy either your own PostgreSQL or your own Superbase. Here it's a matter of taste. Today, I would like to provide a connection that will be maximally understandable to everyone in terms of tools, so we use Airtable. So, ah, Lev Stepanov, this is super. I have already announced to the client that I will go to study automation in the fall. The idea is to parse news articles about motorcycles from around the world using articles. We will create a page on the website for new articles. Ah, good idea, yes, and in general, at the moment, this is a maximally understandable connection. It can be implemented quite simply using ready-made parsers. But yes, and connect it to the website through N8N. You can even set up an agent that will rewrite these news and maintain a full-fledged blog. There are such cases too. There are many of them on YouTube, and on the internet, and everywhere. Ah, let's move on to N8N. Ah, I'll say right away, I will use a self-hosted solution. If you don't have it, or don't yet understand how to do all this, you can use the cloud solution from, ah, from, respectively, N8N. So. By the way, the broadcast needs to be slightly adjusted. Let me adjust it. Do you want it on the side? No, wait, stop, stop, stop, stop, stop. No, I'll do it myself now. So, to the group. Ah, m, so like this, yes, so that there are no extra tabs. So, share screen, window, and like this. Uh-huh. Yes. Airtable, ah, we will need, respectively, N8N. Well, and OpenAI as part of the node, or rather, as part of N8N. Ah, so, yes, we can use N8N, which is a cloud solution. They have a free period. What's the point? For new OpenAI accounts, there are some tokens given for free, so you can fully check the connection in, so to speak, in action, and then figure out how to deploy your N8N self-hosted. But or just stay subscribed to Lazy Method and the guide, maybe soon there will be a guide on how to set up your own N8N self-hosted and not depend entirely on the cloud version subscription and not depend on the limitations of the cloud version. Ah, but for starting and for repeating today's flow, it will be quite, more than enough. Ah, here's what else we'll need. We'll need an OpenAI key, yes, how to get it, I also want to, ah, post it in the channel a little later, but as part of repeating, just the cloud solution. There are free tokens from OpenAI, so you'll succeed. Ah, yes, I'll make it a bit larger now. Let's start. Ah, so, where do we start? We start with the fact that, ah, in order for us to receive messages from Telegram, from any group, we will need to create a Telegram bot. Telegram bots are created using BotFather. Let's add that now too. So, window. I think the screen can be made a bit larger, because on YouTube, yes, it's a bit, 125%. So, let's start with creating bots. We will need, uh, two bots. The first bot will listen to, respectively, Telegram chats. We will add it to chats with clients. Ah, the second bot is needed for us to simply communicate with it internally and request some specific things, yes, like send me a summary from the client for today. Here. Ah, yes, to create a bot, we press the New Bot command. They also have a full-fledged mini-app, which can be opened by pressing the Open button, but somehow I'm more used to the old way. Ah, so, uh, what, what does it want? It wants us to create a name for our bot. So, Telegram. Mmm. Mmm, so Alexeev, Wibot, let it be so. So, now we need to create a link for our bot, Alexeev. That's it, we got the first token. We can already go and connect it, in fact, to N8N. Ah, I suggest creating the second bot right away, which we will use, you could say, as a console for accessing the message database with the client. Ah, so I'm creating another bot right away. So, ah, let's do it like this. TG assis bot. So, Alexeev TG bot. That's it, now we have two bots. All we need to do is get the tokens, yes, of the first and second, and connect them in N8N. Ah, so I'm disconnecting the Telegram stream and returning to the browser. So, I'll make it a bit larger. Please write how you see it now. And yes, the first thing we will do is create a flow that will listen to chats. So, let's create a trigger, yes? Meaning, any automation starts with a trigger. A trigger is what launches our workflow. In this case, it will be on message. And here we need to connect our bot. I have quite a lot of different ones. Uh, but let's, by the way, yes, ah, create a new one. Here we will need to add the token of the first bot that we created, and it's advisable to name all the credentials that you add with the names of the bots that, ah, with the same names that you have in BotFather. Because if some authentication fails, you can easily understand what bot it is and in which bot you need to, for example, reissue a key. So, the token, rather. So, the first bot is Alexeevbot. Save. We select it here. Credential to connect with AlexFirebot. We check that the trigger is set correctly. Message. Also, if you press F2, you can rename the node for yourself. Our first flow will be responsible for recording, recording the message, ah, into the database. Therefore, the next thing we need to do is create a simple table in Airtable, with which we will work further. Before we do that, let's first create an Airtable node, we need the action to create a record. Let's disconnect them from each other for now. What do we need to do here? Here we also need to create a credential for our, ah, for our, respectively, for our table to access Airtable. To get the API key, yes, we need to go to settings within Airtable. Let's make it larger too. We go to account. There is a section here. And go to developer hub. Ah, in this section, we can create a key that we will use further. We click on create token and here again the Create token button. Ah, here, look, an important, very important part. I won't create a token now, because this account is already connected in N8N. But, first, name it clearly, to whom you are issuing this token. In this case, I usually, ah, write the domain on which I have a specific workflow, so I would call it something like NV8N Fluid FY, yes? Meaning, and I will know that this token belongs to N8N, which is our studio, personal, and so on. The next important point is scope. Scope is the access that you grant with this token, yes? Meaning, there are quite a few different options here, but it's important to choose the scope that suits our tasks. Within our tasks, we need to write, read, and read all bases that exist in Airtable. Ah, where can we look at this? We can look at this here in the section for creating our credential for Airtable. Ah, and here we will see that there is the following scope for your token, data records read, data records write, and schema bases read. Meaning, in our case, you should choose this, this, and this. Ah, the next section is Access. You can grant access either to a specific base, yes, meaning a specific table, or you can grant access to all resources that exist in the account, in this Airtable account, yes? Meaning, I can grant access either to all tables, or to a specific table. Ah, most often I just set all current and future bases. Ah, yes, because, well, this is my personal space, and if I have access to the entire base, it's okay. Ah, if you, for example, work with a client, then you can ask him to grant access only to a specific base with which you will work. Ah, and click create token. That's it, the issuance of the token for Airtable is finished. In total, we have already connected two Telegram bots to N8N and Airtable. So, close. Now, now we go to tables, click create. So, I have building app here. Builden your own. Workspace. Workspace. Here. Ah, yes, we create an empty base, let's call it TG Base for Wibot. And we need to configure our database, where our messages will be recorded. I'll make it a bit smaller, it's all too large for me. Ah, so. Ah, what will we record in the base? We will record the following fields in the base. First, name. Here we will pass the Telegram username of who wrote it. Next. Ah, we will pass the message, who wrote it. Save. We can delete all other fields, we don't need them. and create one more new field, which will be called Date. Here it is important to choose the ISO format, yes? This is important later when working with the agent, so that the date is always in the format we need. Ah, we have prepared the table. The next step, what do we need to do? So, I'm thinking, maybe I should just open web Telegram. I'm thinking how best to show this, honestly. Ah, so, okay, let's go back to the flow. And I suggest building the flow step by step first and then checking the operation for receiving messages later. So, ah, what do we need to add? In N8N, there are nodes that allow, uh, so to speak, technical nodes that allow setting certain rules. In this case, since we can receive both voice and text messages, we will need to build two flows. The first one will transcribe and record something into our Airtable, and the second one will record the message into Airtable directly without transcription. Ah, so Tatyana asks: "And where are you doing this?" Specifically now, I am in N8N. Let's continue. We need to create two rules. But in order for us to set up these rules, we need, of course, to receive two messages. At least to receive them, we will have to write to the bot or add it to a group and write in the group. Therefore, I activate the flow to activate our trigger. And now I will create a new group in Telegram. So, ah, I will add to this group, in fact, the bot that we created, the first one, yes, the one for listening. And I will add one more person to check how our entire history is recorded. So, ah, let me show Telegram now. In fact, our group. In the group, there is me, Pasha, and our bot. Ah, now we need to send a few messages, yes, in order to get incoming data, so that we have something to work with, yes? Meaning, we must receive some JSON from the Telegram node in order to then set up the rules for, ah, first, routing, yes, meaning, what we have, ah, the first part, if the message is voice, then it goes to the flow, ah, with voice recognition, yes, meaning Whisper AI from Open, from ChatGPT, and respectively, the second flow, which goes purely directly to Airtable. Ah, one more thing, before we do this, we also need, ah, now our bot will only read messages in which we tag it, yes? Meaning, if I write like this now, the message will come. But if I just write something, this message will not go to N8N. How to fix this? Ah, let me show this too. And here, by the way, we will need, ah, we will need the mini-app that is available for managing all this. So. How can I share it, damn it. Open delete share screen. Voila, finally. To stage. So, here, here are our bots. Alexeev TG bot - this is the bot with which we will interact to request summaries. And we have Wibot, ah, which will listen to our messages. Ah, yes, to get into this mini-app in BotFather, just press the Open button to the left of the text input. Ah, go to bot Settings. And here we need to, uh, change group privacy, yes, meaning we need to disable it, yes, to receive messages from any people who write within this group. Here. Therefore, this point also needs to be done. Ah, that's it, we don't need this thing anymore. We return to our newly created group. So, share screen. And now we can, ah, what to do? We can just take and write something, first of all. Ah, and the second thing I will do is record a voice message in this group. 1 2 3 4 5. Why did I do this? In order to get the starting data for my workflow. So. Ah, let's return to N8N. And if we go to the Executions section, we will see that here we have Executions - these are completed processes that, respectively, have worked. Ah, the first thing we received, we received my text message, yes? Here is all the information about who sent it, why it was sent, why it was sent. There is a date, true, in seconds. We will convert it, of course. And we have my voice message, yes? Meaning, there is my voice message, and there is a file ID that we can download. Now let's quickly build two different flows. To do this, we need to take the starting data from Executions. Let's first build the flow with voice transcription, yes? Here we click the copy button. Uh, and this button will only be available if we have JSON selected. We copied this JSON. Now we go to the first node, click on the pencil here and paste our starting data that we received. So, now we can set up our routing. The first thing we do, ah, here, if you look at the structure, it is as follows, yes? Meaning, message and then some additional fields inside. Ah, how is a voice message different from a text message? It is different in that in the body, uh, in the JSON, respectively, there is a voice object. What should we do? We just need to check if it exists. And then we direct our processing to the flow with voice recognition. Ah, yes. Meaning, I just take and drag the voice to the corresponding field, yes, routing rules, it's the first one. Let's make it a bit wider, perhaps, because of the increased scale, it got compressed. Yes, I set the condition that the object
There exists, and that's it. And now my processing, it will go through the zero flow, and we need to create it. What do we need to do? First of all, since we don't have a voice file, yes, all we have at the start is just an ID file, we need to get this file. Therefore, where our zero comes out, we write Telegram. Telegram, uh, get a file actions. And here, first of all, we need to execute this node. so that the data jumps to the next node and get the file, or rather, substitute the file ID into our GetFile node. Execute this node. Mmm, like this, yes. Let's check that the creds are from the correct bot. And in the response message from our Telegram node, we will get an output, yes, that's the data, yes, we can already download it. We don't download it, we continue building the flow. Here we will need open to transcribe our voice message. Select Openi transcribe transcribe, respectively, recording. And we don't change anything here. Everything needed is already here. input data field name data execute step message said 1 2 3 4 5. Okay. And now we need to write this to Air Table. Connect Air Table. What do we need? First, select the correct credentials from the desired account. Select the base we will work with, TG Base for Wirebot, and select the table we will work with. Let's wait a bit. Table one. By the way, you know what else I would add? This is a teaser for scaling this connection. I would add a single line here and call it chat ID. Uh-huh. So, and here I will update, yes? As soon as we add new fields or change the name in the Vir table, NVOmn understands it immediately and suggests, you saw here, yes, I clicked the button, it suggests updating the fields. So, uh, now in general, what are we doing? Now we need to collect all the information and transfer it to Air Table. We start with a switch node. What will we get from it? We will get the chat ID from it. We will get the username from it. We will get the message from opene. We will also get the date from the switch. Here it is, but we will need to convert it. And we can convert it directly within the fields, yes, there is such a thing as an expression. If we open it, you will see anything inside curly braces is JavaScript. That is, we can use JavaScript in nodes for, respectively, converting some values and so on. And I already have a template here. We will also give this. The only thing we need to adjust a little here. So, what is our field called? So, expression so. Ah, pa-pa-pam. Okay, we'll do it differently now. So, new date we need. New date. Open parentheses, multiply by 1,000. So, as we see, using JSON conversion, we multiplied by 1,000 and got some date. It's important to get the date in a specific format here. And I will do that right now. ISO string. Now I am getting the entire date with time. I will probably throw this thing into the chat, and you save it. So, let's check. Execute step. So, chat ID. can accept the provided value. It says that the value I am passing to chat ID does not match what is needed. So, we can try, you know what to do? We can change chat ID to number. Save execute step. Date cannot accept the provided value. It says the date is not suitable. What if we trim the time a bit more here? Great, guys, we won't pass the time. We will pass zeros instead. I've thrown the updated expression for the date field into the chat. If you don't change the node name, then the expression I've thrown in will fit perfectly and work for you. So, let's check. A value appeared in the table, the date appeared, the chat ID appeared. The only thing is, I don't quite need it. So. So, I need the value without dots, without save. Well, this value will suit me. I'll say why it's needed at the end. So, we're a bit delayed, but now everything will go faster. So, let's update just in case. Everything is working, the record is being created, everything is great. So, now, what do I want to do? Let me send another voice message and see what falls here. Hello. I need a website on Webflow in 3 days. Magic, nothing else. Date is there, chat ID is there, message is transcribed. Everything, guys, how do you like this thing? Well, so far it's just a database, right? So, all the salt will be a bit later. So, the second node. What, what should we specify in the switch node for creating the second flow? First of all, we need to get the starting data. with a text message. Select JSON. Go to the editor. Change it here, click on the pencil, insert the text message. And here's what I'll do? I'll select text. And again, a string value exists. Yes, if it exists, then we go down path number one. Ah, I copy this whole thing. and put it here. Here I need to replace, first of all, I need to execute this node. Here I need to replace message with text. Everything else, everything else is already configured. Click save. And I'll write another message. I need it very quickly in 2 days. Let's check. I need it very quickly. In 2 days, everything appeared. So, that's it, we're done with the first flow. With the first bot, everything is working, everything is great. Now we can create, or rather, add this bot to a specific chat with a specific client. All this will be collected here. And then you can access this database and retrieve some information. What I wanted to say? About the chat. You can upgrade this thing a bit. About this, by the way, about how to upgrade this processing and get the ability to add this bot not only to one specific chat, but to all chats with clients. I also want to release a post on the Telegram channel. But the point is, yes, we created the chat ID precisely for this. We will need to refine this thing a bit. We won't have time to do it within the compressed broadcast, but yes, such a possibility will exist, and this bot can be used in absolutely different chats. As soon as a message comes, it will be checked, what is it? What is the chat ID? If the chat ID exists, then it will be recorded in the context of the same, in the same, respectively, in the same chat ID. If the chat ID does not exist, yes, you have added our bot to some other client chat, then a corresponding chat ID signature will be created, respectively. And, then the agent who will work, whom we are creating now, will distinguish chats. And this is also an important thing that can be done. For now, let's assemble the second flow. It's simpler, it only has three nodes. What will it do? Here we need the message Telegram trigger on message again. Here we should choose another bot. So, we choose another bot. So, token bot. this is our assistant bot, which will respectively form those very summaries for chats. Again, it is activated by a message, yes, by your request. So, here I will do it like this right away. Now we need an agent. We need to create an agent that will work with all of this. Again, in order for us to start working with something, we need starting data. Let's disable it for now, so as not to break the logic. And let's go and write to this bot. Hello. Hello. Go to Executions and get the new starting data for the new bot. As we can see, the node has worked, it has accepted the starting data. Go to the editor and write save here. So, uh-huh. AI agent consists of three components. First is the chat model. Here we again select the model that will work for us. within the cloud solution, the free tariff will have access to 4o min. That's it. In general, it's okay for testing, but for real tasks, I still recommend using 4o. It's more powerful, faster, yes, more expensive, but we are not a corporation here, yes, we don't have thousands of requests per minute, so it doesn't matter to us which model will work. It will still be inexpensive for us. Put $5-10 on Open AI AP, and it will last you for 3 months, even if you use the most powerful models they have. Next, the next component is Memory. What is Memory? It's memory. memory so that your agent can maintain a dialogue, yes, and not lose context. Here, within this task, I think up to twenty messages is more than enough. That is, over twenty of your requests, the bot will fully maintain context. On the twenty-first, it will forget who you are. Well, I'm joking, but nevertheless, it works exactly like this. And the third component of the agent is tools. Tools can be different. It can be some kind of vector database, yes, for the RAG system, it is built on this basis. There are many standard native tools. Air table from the familiar, so to speak. There is, I think, deeple, notion, Discord, Dropbox. What does tools mean? It means that the AI agent can go to Air Table and bring something back to you. This is very convenient precisely in the context of solving our task. We need an Air Table Tool. We need our AIG to go to our Air table, which we created beforehand here, and get the necessary data for specific dates. That's why, what I do? I again check the credentials to make sure it's what I need, resource record. Here I will put search, because I need to search for records. Here I will choose TGBAS for Wirebot. Here I will choose the table we are working with. And the most interesting thing, the most interesting thing is this thing. Filter BYUYformula. I'll say right away, it's not a simple thing. Not simple. Why? Simply because Airtable apparently wanted it that way. But any filtering works through specific formulas. I will also throw in the formula. What's the point? Yes, so I say that most often, at least in my context of tasks, I need correspondence for a specific date and so on, yes? That is, I send my agent to collect a summary, for example, for today. This is the most common case. So, what I say, I want to filter records by dates. The date will be formed by our AI agent. Here you can see the signature from AI date. And, respectively, yes, with such a formula, we will get all records for a specific date that we ask the bot to retrieve. So, what else do we need to do? Any agent, if it, respectively, does not have any, if it does not have any pre-settings and configurations, it will do all sorts of nonsense. Therefore, we must always use a system message, it is an integral part of any prompt, yes, when we give context and understanding to our agent, what he should be doing here at all. And here again, I have a template, and how it looks at all, yes? So, you are a project analyst assistant, the user asks a question or command KP Roadmap Backlog. You can find the client's correspondence data for a specific date in the Tool Air Table. Here you can put two for accuracy. If the user says today, here is my small clarification from the point of view that he was feeding me incorrect dates. In the first iteration, he tried to find correspondence for 2023, which was not very good. Ah, respectively, yes, so we need to tell our bot, our agent, more precisely, what exactly it should do. Here you can also form, uh, how the agent will respond to you. And this is also an important moment. So, all this is okay. Close it. Uh, what else do we need to do? First, I will throw you something that will allow filtering by dates into the chat. Save it too, definitely. Uh, we need to distribute the fields and make it so that our AI agent responds back to the same assistant bot. So, just a second, a small commercial break. Pasha, can you connect and ask the guys. Guys, I'm here. Ah, so I see that our chat is not active at all, but there are people with us. Guys, tell me, how difficult does it seem to you or not? Is everything clear? Maybe someone is already checking live, trying to assemble the same thing, is it working for you? Write your first impression. Perhaps many guys here are encountering N8 MN for the first time, and it seems to you, damn, it's some kind of volatile machine. I don't understand anything yet. Ah, but it's very interesting. Please write, it's important for us to know your opinion. Ah, so my video just twitched. Listen, nothing should have happened. Well, it seems okay. Meanwhile, I see that in the question, how often do you use AI in your work, we have 42% every day. And actually, this is very expected, because I think chat GPT is already used, well, I just use it in all everyday tasks, not related to work. But I think most people do too. Ah, yes, Alexey, did I understand correctly that this assembly can be tried for free? Yes, Kolya said at the beginning that, well, first of all, OpenI gives you tokens plus, as far as I know, N8N, you can also do everything for free. That's why, yes, it's possible for free. Ah, Dmitry writes: "I did something similar recently, but I used Superbase." Yes, you can use Superbase, but here, since it's a training format for simplicity, ah, as Dima wrote above, you can also use another solution, and we just use a table, so that it's easier for you to get into all this, of course, you can use it. Yes, you're here? Yes, I'm back. Ah, I see N8N for the first time. Well, actually, this is probably the main tool that allows you to connect various services, yes. As we can see, we haven't done anything super custom here. All the nodes that we use, nodes are these cubes. They are all already under the hood of N8N. All we need is to make the connection, as we connected Telegram bots and Airtable. Ah, well, how to repeat this? Actually, it's quite easy to repeat. Plus, a post will be released, and you can write your questions in the comments, and I will come in and answer. If something doesn't work for someone, I will definitely help. Well, let's continue. Our timings are already pressing, actually. So, I'd like to speed up a bit. We stopped at the point where we assembled the Air Table Tools, yes, with the correct formula. Next, we need to set the fields, yes? That is, at the current moment, we have the standard JSON Chat, input fields, and this history doesn't suit us. We need to select define below, yes, and prompt user message we need to pass ourselves directly from the node, yes, this will be the incoming message that we write to our assistant bot in Telegram. Next, that's not all. The second thing we need to do is to pass the chat ID into the agent's memory. Yes, again, select define below. Pass the chat ID like this. Message chat ID. Not chat, like this. So that it maintains context within our chat and doesn't lose it. So, uh, uh, mk time production. Ah, so, it suggests deactivating it. Let's check what our, how can I help you today. Ah, yes. So, we've passed the fields. It didn't access Tools because we didn't have any requests. Ah, yes. Now we need to form the last node, which will send the response message back to our bot, yes? So, this whole thing, how can I help you, should be passed back to our Telegram bot. And here we select send message, we pull the chat ID from our, from our first node. And we pull the output from the AI agent. What else needs to be done here? We need to remove the appendix N8N, yes? That is, this is a signature, like message sent from N8N. I usually disable this always. And it is disabled exactly in this way. So, let me try to pass it back. Yes, not those. Yes, everything, the response came back to me. Click save. Ah, well, that's it. Now I'll activate this thing. Send me the summary of the correspondence for today. We send the request to our assistant and see what's happening there. So. I received a response. And what do I see? I'll go to Executions. I won't demonstrate Telegram anymore. Well, okay, now. In response, I received. Here's the summary of the correspondence for today. The main goal is to create a website on the Webflow platform. Client problems and needs. It is necessary to complete the website development in a very tight timeframe of 2-3 days. Proposed solution: it is necessary to meet urgently within limited deadlines. However, details and proposals for temporary implementation are not specified. In short, there's a slight lack of context here, yes? That is, if this were a real correspondence, then you would get something like this. Now I'll just show an example. So, TG TG WB test. No. Ah, damn, now I'll demonstrate my assistant. Stop screen screen. Telegram window share. Ah, so, respectively, for August 6th, the project goal is to create a modern functional landing page, including six-seven screens. Main tasks: development of adaptive design based on Figma mockups, connection of an interactive map with points via API, and integration of Telegram functions. That is, as soon as there is full context, yes, we can sit with Pasha and chat, but that will take time again. But with exactly these settings that we have made today, with exactly that system prompt, you will get such a detailed summary of the project, yes? So, basic development and layout 90,000 rubles. All prices are fixed here, payment terms are fixed here, additional provisions are fixed here, that, ah, the cost includes one round of revisions. All additional revisions are discussed and calculated separately. That is, in fact, you can already transfer this to Notion and send it to the client as a full commercial offer. Well, you must admit that it's worth sitting down for 1-2 evenings on the weekend, collecting these two simple schemes, and freeing yourself from the work of Oh, I didn't show it. Well, I'll show it again. freeing yourself from collecting some data, collecting information from Telegram chats with clients, yes. Here I'll show the project execution stages again. Once more. Project goal: main tasks, solutions, project execution stages, start of work, development stages 3 weeks, five final testing and project delivery, project cost, payment terms, additional provisions. And, ah, here, respectively, I asked it, I think, yes, now it will package it into a full commercial offer. And it took all this from my database, respectively, correspondence with the client for today. Here's such a thing. So. M. Guys, do you have any questions now? Maybe? Ah, I'm ready to answer them a little. Yes, maybe someone is already trying to assemble it, and questions arise in the process, don't hesitate to write. Well. And we will continue, yes, actually, regarding the fact that yes, now exactly this assembly can only be added to one chat with one client. Ah, but the chat ID was created, I repeat, not just like that, yes? That is, we can tell the agent that it should distinguish not only by date, but also by chat ID. And, respectively, then your AIG can be added to absolutely different chats, but the database for all messages will be one. Then, also through the assistant bot, which you create additionally, you just pull the information you need. correspondence for today, for tomorrow, for the day before yesterday, and so on with a specific client. That's why this can be further refined and applied on an industrial scale, adding it to all chats and so on. So, that's it, the practical part is finished. Ah, but I want to tell you about the upcoming, upcoming project, yes, within the lazy method. And it will be a small, small product. Today we talked about the fact that we really have time eaters, we really have a lot of different tools, and we need to figure out all of this. And we decided to make such a small mini-product, which, first of all, contains a lot of value within itself. And basically, all these problems that we discussed today, some have already noted that it's me, ah, all this is solved by one, literally, ah, small product of five lessons. Ah, so, ah, what is the main feature here? First of all, if you don't understand why and how artificial intelligence can be used, you don't understand what, respectively, time eaters you have, you don't understand the abundance of AI tools, then this product can solve all of this for you. It will suit absolutely any specialist, an expert in absolutely any niche, yes, designers, layout designers, marketers, SMM specialists, entrepreneurs. product owners, analysts in general, yes? That is, it is quite extensive and large-scale. I think designers, guys, you know, that often a lot of various contractors are involved for seemingly small tasks, 3D graphics, then, generation of photo and video. Imagine if you could close all of this yourself. And with layout designers, I think everything is clear. We generate JavaScript, integrate it into projects, and so on. Well, and marketers, entrepreneurs, they have more or less similar tasks. Let me quickly show you what this module consists of, yes, why it's called a module? Because it's part of a large program aimed at automating processes in professional activities, yes? That is, we are not talking about some client history here, we are talking about our personal, about what we want to automate specifically, what we want to get rid of, and how with all this to earn more, simply because we will have more time. That's it. Yes. What does this module consist of? The module is quite, it seems like five lessons, yes, but in general, it has everything you need to get started in this field and solve those very problems that we discussed at the beginning. First of all, it's that AI is inevitable in general. We generally discussed this today, but in simple terms, how artificial intelligence works without any super technical details. Key trends that exist at the moment, cases of professionals, yes, so how people are already automating their work. There are an incredible number of posts on Reddit where guys do some, you know, non-obvious things, but so simple with artificial intelligence that it's enviable. So, basically, let's get into artificial intelligence. Then we will immediately proceed to conduct an audit of your workday, yes, and hunt for time eaters. what I talked about today, about the assessment system and the audit methodology for 2024, which will allow us to find in your specific processes what is eating your time, what demoralizes you, what prevents you from developing and moving forward. Ah, this lesson will be entirely, respectively, about this landscape of tools, yes, so tasks can be different, so to speak, in terms of text, visual, voice, specialized, no code, and so on. How to choose tools specifically for me, yes? That is, here we come out with a result. This is a personal process map with potential for AI optimization, yes? That is, we clearly know what we can automate in our activities, and then we map our own tools onto this personal map, well, which suit me, I mean. Well, and of course, we will talk about budgets in terms of time and tools, so as not to automate for the sake of automation, yes, and spend, you know, there's a joke, I spent 10 hours automating something that takes me 5 minutes, so as not to get into such a situation, we will calculate both the time budget and the tool budget here, so that it's all as inexpensive as possible, ah, and so on, yes, so that we don't pay 10,000 thousand a month just for various AI tools. Ah, well, and of course, a personal AI transformation plan, yes, that is, here it will be a full-fledged, probably a workshop even, ah, we will talk about the results of what you have analyzed within the time audit. We will talk about task prioritization, we will talk about priority areas of automation, yes, that is, exactly what you need to automate here and now, so that you don't go into things that, at the current period, either take little time, or don't demoralize you. Well, in short, you don't spend a lot of effort and resources on it. Ah, so, this is what the module looks like. This is the first module of a large product. Accordingly, our task here is to get into all of this. Ah, yes, on the website, you can fully familiarize yourself with all of this in detail. And the most important thing is the price. The price today is super democratic. 3,652 rubles. For the first participants, until August 15th, we have a discount promo code of 56%. Promo code Early Bird. And here you can fill out the form to leave an application. The form is quite simple. Ah, so, this is what awaits you. This is what awaits you today, you could say. Ah, where, yes, go to the website, you will familiarize yourself with all the content that is provided here. Read everything carefully. But the main task, all the same, yes, is to give experts and professionals the opportunity to save time on unnecessary tasks and on tasks that simply drive us crazy. Ah, and we would like to do something else, something more creative, something more, profitable, so to speak, yes, that brings concrete results, and not some operational activity. And the task of this module is to figure out how we can do this. Therefore, welcome to lazy method. Let me add that the program start is currently scheduled for August 18th. How much time do we estimate for this module? Probably a week to complete it. Yes, in a week you can complete everything. That is, on August 18th, you will have access. One lesson. The lessons will not be long, I'm warning you right away, and they won't be nauseating in terms of, like, we'll load you with theory and that's it. No, we'll analyze everything with real examples, yes, just like it was always with me in the TP V Flow courses and all the others, as easily as possible, natively, with real-life examples, real people, so to speak, we will analyze tasks. And 5 days is more than enough to go through this entire program from beginning to end. Yes, I also saw questions in support about whether there will be a restriction to the lessons or not. For the first cohort, we will make the lessons without access restrictions, so there's no point in postponing. You understand yourself that the world is changing a lot. Every day, as Kolya said at the beginning, there are more and more neural networks. Therefore, by postponing now, you are simply, well, you might miss this, I don't know, train, be late for it and, well, not that it will be too late later, but in any case, those who didn't start using computers at first, they still learned later. But the question is, how old will you be when you learn to use all this? That's also important. Yes. And here, in general, this program was born not out of nowhere, it was born from, so to speak, communication with colleagues, yes, and understanding that, well, we are all in the digital sphere and we are all so cool, we do Webflow layout, we use NVOmn, no code, cool, design, startups, and so on. But, damn, I realized that even in my circle, quite a lot of people know that artificial intelligence exists. Many use chat GPT for some of their tasks, but how to systematize all of this, few people know. That is, there is no understanding of how I can integrate all this into my processes. And precisely this first module fully answers all these questions. And how, yes, technically? That's the next story, that's the next step, which we will also tell about someday. But we want to start with such a gradual immersion into all of this. And the most important thing is that we will not give, so to speak, templates, conditionally, this is a maximally personalized story, because everything will be built within this module, yes, everything will be built precisely on what takes your time and what you can specifically automate in your processes and where you can apply artificial intelligence. And this is probably a super killer feature, because there are an incredible number of AI courses, but they mainly give theory on how to write a prompt, that, like, the course is for chat GPT and all that. A practical, practical orientation, probably almost, well, I've seen only two products, perhaps, where there is something more or less adequate in terms of implementation. How do I implement this in my activities? And this is the main question. And 80% of educational products do not answer this. And here we want to completely solve this problem. Therefore, there is such personalization. So, I think we will start wrapping up. Guys, if you have any questions about the connection that Kolya showed today, or maybe you have questions about the training, then welcome to the chat. We'll sit with you for a couple more minutes, answer. And if there are no questions, then we will finish, and thank everyone for coming to our broadcast. Tu-tu-tu-tu-tu. Yes, questions. Our broadcast is a bit delayed, so maybe questions will come in now. Write in general, how do you like this broadcast format? We are planning to do this more often, possibly once a week, similar broadcasts where we will teach you something. Do you like it, should we do this? What do you think? Ah, yes, by the way, the recording will be available for 3 days, as far as I remember. Hello-hello-hello. Yes, recording for 3 days. Yes, yes, recording for 3 days. Yes, recording for 3 days, so don't postpone, don't postpone the assembly. Well, and respectively, yes, I'm waiting for your comments in my Telegram channel. If you have any questions about this automation, then yes. Ah, yes, we will now post a post in Kolya's channel where you can ask these questions in the comments if anything. I see a question from Lev. Did I miss something? This is the first module of a large program. Ah, yes, yes, yes, absolutely so. It is planned, probably, that the main idea is to gradually immerse yourself in all of this, yes, we understand that it is quite difficult to consume a large volume of information quickly. Ah, so, yes, there will be a large program with six modules in terms of not only identifying some, not only self-diagnosis, but also a complete technical solution to your problems. Plus, in the large program, there will probably be something that is not available anywhere else. The fourth module, you could say, will be completely online, yes? That is, it will be, these will be, you know, full-fledged Zoom sessions in which we will analyze your, your automations, yes? You can, you come, respectively, to the fourth module with a specific request, and we will analyze your tasks in real-time, I and a second expert in artificial intelligence, and suggest and map automations directly for you within these activities. That's it. Yes, so there will be pre-recorded lessons. These are five modules, including the one you see today. And the fourth module will be completely online, where we will sit in Zoom as a whole group, exchange experience, and everyone will solve, definitely solve, guaranteed 100% solve their problem in terms of optimizing, automating their professional activities. Yes, I will also add here that by purchasing the first module, in any case, even if you don't go further, then you will get benefit for yourself and will be able to identify your time eaters, and possibly eliminate some of them. That's why, yes, so this is a self-sufficient product, the first module out of six. It is self-sufficient in itself, yes? That is, and then with the knowledge that you can get within this module, you can calmly move forward independently. No one cancels that. Yes. Well, I see that we have no more questions. If anything, write to our manager. Let's add the contacts too. Well, if you have general questions about training, you can submit an application. It doesn't obligate you to anything. You can clarify everything that interests you with our manager Anton. Yes. That's it, everyone, thank you very much. Stay subscribed to lazy method, come to our broadcasts and optimize your professional activities. Bye-bye everyone. Bye-bye.