Full Agentic AI Course on Grok Bots - Chief AI Officer Playlist
Publicado no IA Dots ·
Agentes de IAAI SignalAssistente pessoaldot-ai_agentsdot-artificial_intelligencedot-geminidot-llmidioma-enInglêsyt-JFZ5SjGC6bY
Vídeo em Inglês, publicado pelo canal The Cutting Edge School.
About this content: The summary, key points and topics below are generated automatically from the video transcript. They reflect the content, opinions, claims and predictions presented in the video and do not represent a position, validation or fact-check by IA Dots.
Video summary
The video presents a comprehensive tutorial on building and managing an ecosystem of AI agents using Grogbots (powered by xAI's Grock) to execute end-to-end tasks for a direct-to-consumer brand preparing a Diwali promotional offer.
Key points
- The video demonstrates how to set up and deploy specialized Grogbots—such as a chief of staff, research agent, creative designer, website builder, and support agent—to handle tasks for a D2C brand called Healthy Mitai Co.
- The presenter explains the Grog ecosystem, detailing models like Grock 4.7, context windows, tokens, retrieval accuracy, reasoning quality, and API access.
- A virtual computer feature within Grogbots provides a shared workspace and file storage so agents can coordinate and reference previous tasks without repeating information.
- The tutorial covers connecting external applications (such as Gmail, Google Drive, Lovable, 11 Labs, and Apollo) as plugins to enable bots to read, write, and execute workflows across platforms.
- The workflow includes automated B2B and influencer prospect research, custom ad and brochure generation via Hexfield, website creation, and integrating an 11 Labs conversational agent for WhatsApp support.
- The presenter discusses methods for packaging these AI solutions into public templates to sell deliverables or agent setups directly to clients.
Topics
AI AgentsGrogbots EcosystemContext WindowD2C Brand MarketingWorkflow AutomationAI IntegrationsBusiness Operations
What you will learn
- Grogbots operate on a 'done-for-you' persistent model where a chief of staff agent delegates tasks to specialized subordinate bots running in the background.
- A context window determines how many tokens (or words) an AI model can remember and process in a single attempt, impacting how complex a task history can be maintained.
- Assigning one specific job per bot prevents memory overload and keeps instructions clear, making it easier to track, fix, and hand over tasks.
- Automations can be built without code by linking website inquiry forms via webhooks to trigger routines that automatically generate personalized brochures and email drafts.
Questions and answers
What is the minimum subscription plan required to use Grogbots?
Grogbots are included starting from the premium plan, which has a minimum price of $30 a month.
How does a context window affect an AI model?
A context window determines how much information—such as previous conversations, uploaded documents, and system instructions—an AI model can remember and use in a single attempt.
Why does the workflow use separate bots for specific tasks instead of a single bot?
Using one bot per job prevents the context memory from becoming cluttered and overburdened, making each agent easier to track, fix, and manage.
Video transcript
Transcript from available captions; it may contain errors.
This is the only video you will ever need to understand Grock and Grog bots. Welcome to our chief AI officer playlist from the cutting edge school where we will teach you how to think and work like a chief AI officer for either your own company or for your clients. I totally agree with Elon Musk when he tweeted that Grogbot can now help you earn money. In today's course, I will teach you how to build AI agents for a D2C brand called Healthy Mitai Co. and prepare a Diwali offer using a team of AI agents. Now, don't forget to click on the subscribe button so that you can learn from the upcoming episodes as well. So in this video you will learn how to build agent number one to research your competitors. Agent number two to create product images and ads. Agent number three to build a website. Agent number four to handle your customer inquiries. And agent number six to find corporate clients. We'll teach every single thing step by step. If you document your notes in the comment section of this video without using AI, you will [music] stand a chance to win a 24,000 rupees scholarship from the Cutting Edge Academy. Details are in the description. To give you examples, a business owner asked Grogbot to find customers who had left his business in the last 6 months, email them, win them back, and collect their feedback. In fact, someone used Grogbot to negotiate with 40 suppliers overseas on WhatsApp. Now, think of it as having an entire team of AI employees at your fingertips working across apps and websites even when you're away from your computer. So, without further ado, let's start with chapter number one. Now, before we enter Grogbots, I think it's super important to first understand the entire ecosystem of Grog. X AAI is the company that owns Grock and the entire ecosystem. Now this ecosystem includes Grock AI which is a family a group of LLM models that you can chat or reason with. This LLM is the large language model which is the actual AI the intelligent part. Then there is Grock imagine which is used to make images. So how Gemini has nanobanana, chat GPT has GPT images, Grock has Grock imagine. Then we have Grock build. This is almost like claude artifacts or GPT sites where you can create websites and do coding. And then we have Grock bot. This is the latest agentic platform for building your own AI agents. So you could also think of this as Chad GPT dots or Gemini gems. We also have Grock voice and the XAI API which lets developers and solution builders integrate the intelligence of Grock into their own software or voice agents. Grock now has a collection of models. So at the moment that I'm recording this video in October 2026, we have till Grock 4.7. 4.7 is the most important, the most powerful general model everyone can use. We also have some cheaper and faster models as well, but I don't think you need to worry about those right now. Now, Grock 4.7 has what we call a 500,000 token context window. Now, this is a very important detail. So, I hope you have a pen and paper in front of you. Let's first only understand what a context window is. Imagine you hire two AI assistants in your company and you hand both of these assistants 20 books and ask them to find a common lesson between chapter 1 of the first book and chapter 10 of the last book. Now, one assistant can only look at a few chapters at a time. If it reads more than 10 books, it starts to forget things. the other assistant can read and remember all of the information of all the 20 books and then give you a response. That is the basic idea behind a context window. A context window determines how much information any AI model can remember and then also use that in a single attempt. Now this is one of the most important specifications of any AI model because this context window decides how much conversation or how much documentation or how much code can any AI model remember when answering your question. Now why do they call it as a 500,000 token context window? See AI models don't process information or text the way humans read. So humans always read words. In fact, when we read something, we also read in between the lines and in between the words as well. But AI does not only read words. It actually breaks your text and each and every word into smaller smaller pieces. These smaller pieces are called tokens. For example, if AI is reading Elon Musk founded XAI in 2023, that sentence first gets converted into smaller words. Smaller words then get converted into a sequence of tokens. Now, a token might be a complete word or even a part of a word. And as a rough rule for ordinary English, one token is almost 0.75 English words. That's the average that usually comes up. Which means that 100,000 tokens is almost 75,000 words and half a,500,000 tokens is almost 375,000 words. So when Xi says that Grog 4.7 has a half a million token context window, it means that combining the input and the output in a chat for Gro 4.7, the context is across 375,000 words. Now please note that there are other AI models as of this moment that have higher context windows as well. So GPT6 Astra is at 1,ion50,000 token window. Gemini 4 argon is also at around 1 million 48,000 token window. But please note that a model that can read 1 million tokens is not automatically better or understanding or even reasoning. Big context windows do not guarantee better results. A context window is almost like calculating how much information is fitting on your desk. But model intelligence is all about how well can you understand the information which is kept on that desk. And for all of these intelligences, there are different metrics and different evals. EVALs are basically tests that people run to test the intelligence. Now, how exactly is this context window calculated? Many people think that context window is the sum of all the tokens that are there in the input prompt plus all of the tokens in the output prompt. But that is not entirely correct. Your input tokens include many additional things. So in chat GPT in grog you have this feature called custom instructions or system instructions. Then if you're on the paid version of these AI tools, they also have some memory as well. So when you use these AI tools, they understand what you do and your preferences. So all of these additional things like the system instructions, the memory, they also get added as you write your own prompt. So things like uploaded documents and context from previous chat and all of these additional things they also contribute to the input tokens. Now this matters a lot because when it comes to enterprise use cases all of these additional things inflate a lot they bloat a lot. Also a model's maximum response length is a separate specification altogether. For example GPT6 Astra has a 1,50,000 token context window. But guess what? The maximum output limit is only of 128,000 tokens and this keeps on changing. So you cannot ask a GPT6 Astra to produce a 1 million token answer in a single response. So what exactly happens when you use a Grogbot as compared to any other tool and what happens when you ask a Grogbot to do something which is very very complicated over a long duration of time because agents are not just getting a prompt. agents operate for a longer horizon. So let's just say that Grogbot has a certain limit of context window. Obviously at some point it will surpass the context window because it's not unlimited. So what happens when the task becomes so elongated that you end up finishing this context window. Does the AI stop? Does it start making mistakes? Not really. So, Grock and even other AI tools have some very simple solutions to make sure that they don't make mistakes even when the context becomes large. Once the conversation becomes too heavy, Grock can either remove older messages from the active context, summarize previous discussions into shorter notes or retrieve only the relevant parts of my previous information or start a new context with selected key details. Now the AI world has given all of these techniques some really fancy names. So sometimes people call this as truncation summarization retrieval. They have this word called context compaction as well. But in each of these cases, AI results are always a summarization of the previous conversation which means that you can never blindly trust them. AI results always need to be the baseline and not the finish line. So you always and always need to remember the shape framework that we have taught on this YouTube channel since a very long time. SOPdriven human in the loop arbitrage product is no longer the mo and be employable. If you're here for the first time the article is in the description you can read it after this video ends. Basically whatever AI gives you has to be the first draft and not the last draft. So if the context window is not the only thing that determines an AI model's efficiency, how do we accurately judge whether an AI is better than XYZ AI? So there are three important things that people use. First is of course the context capacity which measures how [snorts] much information can fit. But then we have retrieval accuracy which measures how reliably an AI model can find relevant information from a large pool of context. And third is the reasoning quality which measures how accurately or how relevantly it uses that information to solve a problem. Now Grock 4.7 can be either used through chat or if you want to use it in your own apps you can use it via API. What that means is that if I am selling an AI solution and if I want the consumer to get intelligence where do I get that intelligence from? So if I want grock to supply that intelligence, I will access Grock via the API access. Now right now when I say Grock 4.7, you also get access to four reasoning levels. So Grog bots basically require you to also define the level of intelligence you want to be activated. Now what are Grog bots? Grog bots are customized agents that can use Grock 4.7 to complete different things for you. So when you use Grog 4.7 through chat, you have to ask for a solution again and again because you are having a two-way conversation. It's like phone calling a smart person. But in Grog bots, you just have to define a goal and give it access to different tools. You give it security permissions and then the Grogbot micromanages itself. It's almost like having a 24/7 assistant running on its own. So this was just part number one. I wanted to set some context before we actually get into the building part of things. In the next part, part from our cutting edge research and development team will teach you how to build these solutions end to end using Grogbot. We have a practical problem in front of you. And I'm telling you folks, if you watch this entire video end to end, you will know every single thing required from an understanding perspective, from a practical perspective to actually become the chief AI officer in your own company. So without further ado, part the stage is all yours. Thank you for introducing me. Diwali is right around the corner and every D2C brand is racing to reach the Indian homes first. But the problem is an upcoming brand does not have the resources to compete with the bigger legacy brands. Now imagine walking into such a brand like the healthy mitai company or the Indian sweet house and offering to help them with an army of bots that can make their ads, answer customer queries in seconds and generate ads and visuals for them. By the end of the video, you will have a team of bots that does that work for a brand and you approve everything before anything goes out. We will be using Grogbots to build this offering one by one. Grogbots is by SpaceX AI. So every prompt, skill, routine and template that I will be using is in a free agency kit linked below. So you can copy and replicate it for yourself. Now before we build anything, let me show you where you will be spending most of your time. This is Grogbot by SpaceX AI. Let's start downloading it. You have to visit x.A/bot AI/bot and then click on download for Mac OS. If you have a Windows PC, download for Windows. Now, while it is being downloaded, let me show you the pricing of it. Inside the pricing page, we have four plans. The first is free. Then we have plus, premium, and super. Now, now you need to note that Grogbot is only included in the premium plan and the plans above it. So, the minimum price to use Grobots is still $30 a month. Now that the Grogbot is downloaded, let's head over to it. Once you sign in, this is the page that you will land on. Now, since I have been using Grogbots extensively earlier, I have a lot of bots that I've made previously. You can click on any bot and start a conversation with it. And for our tutorial, I'll be hiding all of these bots and creating them from scratch. On the top, you can see Grogbot. So, Grogbot is like your proactive executive assistant. Its task is to keep understanding what you're working with, what your day looks like, and whatever you do, and the tools that you work with. The Grogbot can converse with all the other bots that you create and manage them. It has all the context about you and your work. So, let's go ahead and make our first one. I'll go on the plus icon. I'll name it research. give it a shape and a color and then create it. Now, as soon as I type research, it has taken the context that I want something researched and this will be the job description. So, I'll be using a tool called Whisper Flow to give it its job description. Find out everything about a brand's competitor and run a search every evening at 5 to tell me if any competitor made a move. The brand is Healthy Mitai Company. And that's it. That's a bot. I just gave a job description to an AI bot and now it is writing its own purpose, writing its methodologies and working on it. But you might ask, how is this different from a chatbot? So a chatbot gives you text to copy into other tools. It gives you the outputs. A grogbot on the other hand is like an employee. It uses its own cloud computer, the connected tools, the website and files to get the work done. It keeps going on in the background, coordinates with the other bots as the teammates. So while the chatbots work on a do-it-yourself model, these always on AI agents or persistent AI agents work on a done for you model. They are proactive and they observe what your day is like, what tools do you use and then they themselves try to get the work done and hand it to you. These bots aren't like the air tools that we used earlier. It's like hiring a workforce of interns with a manager. So you delegate the work to the manager and then the manager figures out who to delegate the work further to. The AI bot then decides what's the best tool as well as the shortest route possible to get the work done. This way the tools will keep changing but your workforce stays the same. You create the workforce and then they keep updating their skills based on the new tools that come in the market. Now it says the healthy mid high competitor report is done. It says that there is a brand which is the biggest threat. It has given me its findings and also a mdown file. But now you'll be probably asking why did I create a research bot? Why only one bot for one task while I could have given the same task to the grogbot. So why not use one bot for everything? Because every bot has its memory and instructions built up around the job that you give it. Here comes the idea of context. Whenever you give a job description or a task to an agent, it gets stored in the context memory. So whenever you give multiple types of task to the same agent, the context memory gets added up and whenever you ask a question of a specific task, the whole context memory is revised by the agent again and again to answer a very small question. This way it eats up inside your usage because every time the same context memory is loaded again and again which overburdens the agent. So if you put ad briefs, customer complaints and invoices into a single bot, it gets messy. It's like overburdening a single employee to do five jobs. A single focus bot is easy to track, easy to fix, and easy to hand over to a client. So our rule is simple. One bot, one job. And a bot can only get as smart as the tools it can use. So you need to connect your tools like the Gmail, notion, Google Drive or Shopify or anything that you work with in a daily life. And once the tool is connected, your bot can read from it and work in it for you. Now in order to connect any tool, you simply have to go to connect apps, click on it and all the tools get available to you. So you have recommended tools, password managers, productivity tools, communication tools, design, code, data, etc. And according to the latest update, now we can connect superbase as well inside Grogbot. And finally, when you click on the bot's description, you have details, library, and computer. What you see on the left is the details. Under details, you have the routines. Routine is a job that your board does on its own. it's on a schedule or when something happens you can just pause any routine by clicking this toggle off and the same routine pauses. You can again resume it and it starts on the same time. It is triggered on the same time. Routines is how your bots keep working when you are not. So now three important habits before we start. One, give every bot a clear job description. A vague brief gets vague work. Second, connect only the apps that you want your boss to use. Third, you approve anything that goes out to a real person. The bot drafts, you press send. So, here's how this will work. My bots will do the routine work. I will handle the judgment as well as the exceptions and the human the client coordinates with. So, if you run a D2C brand or want to serve one, this is that setup. Start to finish. Research, outreach, creatives containing ads and visuals, a store page, customer support, orders, and reports. This playbook fits any D2C brand. So let's build a team that will be working on all these segments one by one. Now the first step I'll do is hire my chief of staff. This is the bot that is the project manager. I'll be directly conversing with it giving it the overall task and it will be responsible to hire the right bots and get the work done by them. So let's get started. I'll go on the plus icon. I'll name the bot the chief. Give it a color and then create. We are working with the healthy mitai company and you are my project manager. I will be directly coordinating with you and whatever tasks come to my mind, I'll tell you. It's your job to hire the right bots and get the work done by them. I will just tell you the outcome and you will figure out everything that happens in between. You will be coordinating with them, verifying their outputs and instructing them whenever you need to. You should also create a group chat for all the bots that are working under you so that I can just come to that chat and allocate the job or request information whenever I need to. That way we can all work together in a single workspace. We are trying to offer a wide range of services to the healthy mitai company including media and asset generation for example images, videos, ads, brochures, slide decks and so on. And since they already have a website we'll try to make a website on lovable that is much better than this. And thirdly, we'll be setting up a customer support agent on 11 labs that is connected to a WhatsApp number so that whenever a customer sends a query, it is automatically answered in seconds directly on WhatsApp. Moreover, you will be handling the outreach. So, you need to shortlist the right clients. You need to find out the emails using Apollo and then send personalized outreach mails to them along with the research text as well as the personalized brochures. Let's send them the prompt and wait for the output. Now, while the chief of staff is getting set up, let me show you a very special feature of Grog Bots. This is the computer. Now, right now, the computer is asleep. Let's try to wake it up. This is a virtual computer that all the boards have access to. So, when I open the chief's computer, I can see three icons. one is Chrome, one is the code in terminal and third is the file storage. Let's take control of this computer. When I'm inside it, I just open the file manager and I can see all the files and folders that have been saved. So in whichever bot you're working, everything gets stored in the workspace inside this folder. So that every bot has the reference of what you're working on. This way the context gets shared among them and you don't have to repeat yourself again and again just like you would have to when you're opening a new chat inside a chatbot. So let's close this computer and get back to the chief. The chief has come back with an output as well as a few questions. It has also found out real gaps in the website which it thinks it can fix. So here our judgment implies that we need to check if it can actually do it or not. And it has also set up a group chat. When I click on this, I can see that this is the healthy mitai team. And inside when I go to the members, I can see we have the chief, the designer, the writer, the proposal creator, the research, and a bot called friend. So somehow it has picked the reference of one of my old bots, which I don't require. So I'll just ask the chief what it is about. But now instead of texting, let's try calling up the chief. I click on this button. It's called a live voice chat. Hi chief, I have seen the group chat that you have made. I kind of can't figure out why you have added some old bots. I just want you to create new ones for specifically the healthy mitai company. And this group chat will be specifically for the healthy mitai company project. The group name is healthy mitai with a H, but you somehow haven't added it. And thirdly about the outreach question. Yeah, right now we are doing the B2B outreach for the healthy tai company but eventually after we are done with that we also be doing outreach to other brands who we can cater to. That will be done in the last. Right now we are figuring out how to create all these services for the healthy matai company. So could you please answer them and let me know. >> Three fixes coming. Renaming the group to healthy matai with an H. swapping those old bots for fresh ones built for this project and locking the outreach orders so healthy Matai B2B comes first with other brands after that's the job running now want me to ping you the moment the new bot lineup is settled >> perfect yeah let's do that and do you have any suggestions for me on the way forward or do you think that there's an immediate step that we should start with >> since the research is already done the natural first move is mapping who the B2B buyers actually are cloud kitchens, corporate gifting teams, health focused retailers in Pune. So the outreach has somewhere real to land. >> I think let's just focus on the healthy mai company for now and then start creating these services and assets for them. I want you to delegate the work to all the agents that you have talked to and you have hired. I can see that you have hired the HM website, HM creative, the support agent and so on. So let's get started. I have connected all the apps to you already and I want you to start working on the deliverables and get them sorted as soon as possible. >> Delegating across the whole healthy matai roster now website creative support the rest. >> Awesome. See you later. Catch you soon, An. >> So this account belongs to An and that is why it is referring to me as An. But did you just see that I just had a conversation with the chief of staff for my project and it even cut the call in the last by itself. And on the left I can see that it has hired a bunch of specialists that are working on the roles and each of them is telling me the updates they're working on. So the website one is starting to research on the website. It's starting to create a lovable version. The support agent is building the knowledge base for the company and so on. So while these bots are working, let me show you the apps that I have connected already. Now when I go to the connect apps button, I have in productivity I have notion added in communication I have Gmail, Slack, etc. Now X as a connector is available but in the recent updates X is now inbuilt inside these bots. So now whenever you need to search something on X you can just let your Grog bot know and it'll do the research for you across X across all the post or your competitor's post for you at no added cost. Everything is included in the plan. Then for design I can add Canva or Google Slides. So let me go ahead and let's try adding one. As soon as we land on the Google browser, let's login with our email ID and give it access. The authorization is complete and we are back on Grogbots. Inside coding, I have already connected lovable, but if you want, you can add something else as well. And these are the apps that are currently added to it. In Grogbots's terms, these connectors are called plugins. And next comes skills. Now skills are the repeatable processes a bot must follow to get a job done. These are the step-by-step instructions of a single process. And these are very similar to the skills that you use inside cloud chat GPT or so on. And coming back to the main page, you have the feature called bot templates. And when you click on bot templates, this is where you land. This is a marketplace for all the bots that have been created by the grog team as well as the community of people who is using grog bots on a daily basis. So from the grog team, these are the official bots and then they're divided on the basis of their use. So we have engineering bots, sales bots, marketing, design, personal and so on. So I can just go ahead and add any bot that I want. Let's go to personal. I have tech demos by Matt Palmer, chief health officer. So let's go ahead and let's check out the tech demos by Matt Palmer. It has four parts. The memories, skills, routines, and integrations. So memories are the facts that a bot must know. These include the preferences a owner has, the bot owner has or specific facts that you have made it remember. Then you have skills which are playbooks of the processes that you want the bot to run. Then you have routines which are basically automations that run on their own either triggered at a specific time or based on an event. And then you have the integrations or the apps that this bot can use. So on the top we have the bot description which says that it's a weekday xbookmark scout that picks one new library to demo asks for approval then plans and builds it in a sticky monorreo with cursor cloud agent. It's for anyone who wants a daily public tech playground and not a change of digest. And similarly, there are so many BS that you can pick from. So let's go ahead and try downloading one. So this is an AUes on the same bot marketplace. And let's import this bot. The moment we click on import, we'll have to open Grogbot. Inside this, we have a page where we have to add the bot manually. Once you click on that, this bot is now a part of our workforce and I can ask the bot to work with my own tools and it starts setting it up immediately. So while I was explaining these tasks to you, let's head over to the chief because it has come to me with a few updates. It has given me 30 new messages by the time I was explaining these topics to you. So it has assigned the work to all these boards. Research is building a list of 30 or more buyers. designer is making brand kits. Uh HM creative is focused on ad creatives. The website one is working. So it says the brand kit is done. Pulled straight from the live site. It has picked the colors from their website and it is asking me the lovable project it should work on. So I'll just pick one. So it has come up with a few images and a brochure. This brochure is professional and decent but I think it can improve on it. Hi chief, I want you to be using Hicksfield whenever you want to create these assets. All the media assets should be created from Hickfield. All the website development should happen on lovable and the customer support agent should be on 11 labs and let me know whatever your progress is. I also want you to create a Google Drive folder for the healthy metai company where we store all the deliverables in one place and you also need to create a brand bible that that acts as the holy grail for all the words to refer to. This brand bible should contain everything the verified phrases the pricing the items available and the items listed to the website. Everything that the board should know of should be inside this bible and it needs to be very in-depth. But please limit it to under 20 25 pages because I don't want you to spend a lot of time on it. I also want you to check all your outputs once and compare it with the brand and use the actual prices details and phrases they have mentioned on their website. >> Locking in Hicksfield for media, lovable for the site, 11 Labs for support, plus a drive folder for healthy matai and a deep brand bible under 25 pages. all cross-cheed against the site's real prices and phrases. That's in motion now. The folder and Bible are the first things being built. And once the Bible's locked, the other outputs get audited against it. Want me to flag you the moment the Bible draft lands or just let it all come through together? >> Yeah, it's better that you let me know. And do you have any suggestion that you want to add right now? One suggestion, have the Bible include a one-page executive summary up front so the board can get the gist without reading all 25. Everything else stays in the back for when they need depth. >> Perfect. I also want you to use Apollo and shortlist the next 20 people that we can reach out to on the behalf of the healthy mai company to sell the B2B offerings which is a which are the bulk orders that we can sell to their companies. shortlisting 20 B2B prospects in Apollo for healthy Matai's bulk orders. >> Thank you chief. As we close it, the chief has already given me an update of what they have done and each individual bot has also told me on my direct messages. So every asset that you have seen from the chief, the same asset can be seen from the individual bots. Now the reason I'm using 11 labs for the conversational agent is because the core ability of 11 labs is to offer conversational agents with a very good enterprise security. So you need to keep in mind that whenever you are building for any of your clients your apps your bots your agent anything that you build needs to be completely secure because whenever a client's data leak happens you become the data fiduciary. So it is your responsibility to make sure that everything that you send out is secure, safe and verified by you. So right now I have 11 labs handling the conversational agent on WhatsApp. Later I'll be adding the conversational agent to my website. So this way any customer who lands on a website can directly converse with it and get their doubts cleared without them having to call the employees of HMC to clarify their doubts. Inside a plug-in like the level labs you have the accounts. This is where you add the connected account. Then you have the tools. The tools are the actions that your Grogbot can do once it is connected to the app. So inside this it has agents calculate LLM usage, agents delete. So this means that it can delete the agents, it can create new agents. Everything that you can do can be done by the app itself. Then you have the apps. Then you have the skills. Now these are the inbuilt skills that 11 Labs and Grog Bots have compiled into one and you don't have to create any of them. They are already set up for you. The same goes for lovable for creating the websites. as well as hicks field for creating the image and the video assets. So the outreach bot has come up with an excel of all the prospects that it says that healthy mid high company can reach out to. So we have the company name, the segment, the decision maker title, email, email status if it is verified or not. And this is very important because when you have the verified emails, you know that you are talking to the right person. When these are not verified, you can leave them or skip them out. And this is not just for emails. This is for every output that you get from an AI. You need to check if it is actually verified or not. Then you have the LinkedIn ids, the phone numbers, the fit notes, a personalization hook for that person and the suggested deck. So this way it has given me a list of so many people that can reach out to and in so much detail. Now had this been a normal chatbot I would have to take this whole Excel and give it to all the bots. I need to work with it. But right now the same Excel is in the shared memory of all the bots. Every bot can view the Excel and use it as per the requirement. This is the power of hiring a team of bots and not just working on the tool yourself. So when I head back to the chief again 14 plus new messages and it has given me the list of the 20 people that we can outreach to and also the website that it has created. So let's head over to the website it has just created. So the lovable screen opens and it has done some conversation with lovable and this is the website it has finally delivered. So let's just publish it and let's view the website. And this is the website that has been created. It has added the visuals it had created on Hicksfield. It has added the images of the gift boxes. Has listed a few products and images. If let's let's try clicking on some of these. Let's go at this combo. And now we land to the healthy mitai company's official website. So it has connected the two so that you come back to the official website and you can select whatever you want to. Then it has written a section about why healthy mitai. Uh what are the best sellers, how you can buy them, a FAQ section, what is healthy mitai, what are the package that you use and then the contact details in the bottom. Then we have the corporate gifting site. So it has added this tag of FSSI and all the certifications that it has already the reasons to select them. Who do they serve? It has created this section of corporate HR, HR wellness programs, hotel banquetss etc. What is the process and what are the testimonials? Then it has also added an inquiry page. So if you fill something in it'll get stored. I'm not very sure where it is getting stored but we'll be seeing that soon. Then we have the gift boxes and all the products that they have and has added these tags of gen gluten-free gift boxes bestsellers. All the categories have been filtered and added then the contact details have been listed and an inquiry page has been set up. So if you see in front of you everything was handled by the grogbot researching about the brand researching about their products then from those products getting the visuals made then using those visuals and the brand scheme to create the website. Till now we have spoken to just one project manager and all of this has been set up for us. And on the bottom right, I can see the conversational agent has been added. But how did this happen? Why did the chief of staff create this conversational agent without me asking it to add to the website? So here comes the concept of memory. Whenever you tell the grogbot your preferences, there is a very high chance that it will follow it again and again in every similar project that you make. So we have created a lot of websites for a lot of different clients like this and throughout the time that we have spent on Grogbot. This was usually our procedure. Whenever we created a new website for any brand, we used to add a conversational agent from 11 L apps directly to it. So it followed the same preference. But how do you make sure that this process is followed every single time? How do you ensure that? So there comes the concept of skills. Just like the preview of the skills that I had given you earlier, it's time to create a new skill of our own. So let's head back to the chief. While it has given me a few outputs, we'll come to that later. Hi chief, I want you to create a skill such that whenever you create a new website and if there is an agent on the 11 laps present, I want you to link them up such that the website will always be created with that agent present in it. Setting up a standing rule. New websites get built with the 11 Labs agent wired in from the start. >> Show me the skill file that you have just created for this rule. >> That skill is still being written. So there's no file to show yet. Want me to ping you the moment it lands? >> Yeah, please do. Thank you. >> We'll do. >> The task that I've told the chief of staff is to create that skill. Once that is done, I'll see that on my chat. So it says that I have saved this skill which is website and 11 labs agent. It has named that skill that. So when I click on that skill, I can see the name. I can modify that whenever I want to. I can see the description, the instructions, the steps, everything written end to end by the chief. So all the technical details have been added to it directly. And then I have the publish button. So under the button it says that publishing a skill needs a cursor team. The reason for that is that whenever a team of people is working on Grogbot, you can have a skill shared by all of them. So that not only the skill is used by all the agents but all the agents of all your employees can share the same skill and that is how sophisticated this ecosystem has become. Now let's check out the outputs that the chief has sent to me. This is the video. This is a 15-second video. Let's check it out. So it has created a video around all the mitis, all the sweets from the healthy mitai company. It has used the exact images present on their website as well as the typography, the logos, the colors, everything has been added end to end and has also created the brand bible. So I can just go to it and and this is a very detailed bible under 25 pages that I had asked it to and this is only for 14 pages but it covers the brand story the founders the verified phases and tagline it can use the tone styling wording the product catalog they have as listed on the last 15th of October the diet tags matrix ingredients they use the collections that they have the offers They have claims register everything everything is added end to end. Now the part that you need to focus on is it has added these verified and unverified stats. The verified tags are the pieces of information that every bot can use as it is because these have been pulled from the official website. The unverified tags are the derived facts. These are the things that you as the intermediary between the agents and the brand has to get clarified. And I'm just amazed by the detail and the depth it has gone to for creating this crisp document. And on the bottom we have visual dos and don'ts. The brand guidelines, the brand colors, the typography, everything added to the document. Now I can just take this document to the brand and say that this document is enough for you to give to any agency. This itself has become a deliverable for them. Now let's head over to the group chat that we had earlier created and let's see the conversations happening in that. Now the chief has informed all the rest of my bots that the brand Bible is a single source of the truth. The chief has gone ahead and informed all the bots working under it that the brand Bible is the single source of truth. So, let me go ahead and ask my members, what do you think is the next step? Hi guys, what do you think is the next step? And do you think we should be adding any more members to our team? Do you require someone new? Now, while they're answering, let me tell you my thumb rule for creating a new agent. So, whenever you're giving a job description to an agent in that sentence, if there is an and, for example, create images and research. So anything after that and becomes the job description for the next agent. So it is better to have one agent per job as we discussed earlier and create the new agents whenever you hit that and. So my agents are replying. The chief says I don't think we need anyone new right now. The next step is getting these from them which means from the brand which are the B2B prices, delivery timelines, the certificate copies everything. The creative has given me its own suggestions. Website agent has added its own. Same goes for the support agent. So the chief has gone through all of the responses and given me a summary of what every what is saying, what are the deliverables, what it needs for me to give. Now mind you, you can't have a live conversation with the whole group chat. So you only have the text input or the voice input right here. And if you need to talk to a specific agent, you have to go to that specific agents tab and then converse with it. Now that I've saved the skill, you need to keep in mind that a good skill answers six main things. When to use it, what it needs like what are the inputs, the steps, how to check the work before showing me and what to hand back as well as what needs my approval. So next week when I get a different brand, I can use the same skill because the processes remain the same, my preferences remain the same, just the brand inputs change. Now coming back to the group chat, a lot of messages have been received and it turns out the support agent has gone ahead and tested its output again and again because as I told you the grogbot learns from your preferences and through a lot of times I have told it to test its output again and again and that is why the testing has been done and I received the final output. It has even gone ahead and saved the output inside test report.md. So somehow it has taken the number mentioned on the website as the phone number for the customer support agent. But since it does not belong to me, I'll just ask it to use my own number which I will set up manually. And I have asked it to give me the link of the 11 Labs agent as well as the drive folder. Now mind you tools like Hexfield, 11 Labs, Lovable, Apollo, they're not free. They do have some free plans, but we are on their paid plans. So, we have listed the models as well as the plan they're on inside the description so that you can try them out as well. So, the chief agent has given me the drive in which everything is saved. Let's go ahead and check it out. So we have the website containing the list of the improvements it had made. The support agent which has the agent prompt the research it has done till now which has the bias CSV the proposals it has made till now. So it has made some PowerPoint decks some word documents. Let's check one out. So it has created a template for itself to use whenever it's sending out any proposal to any company and that is how the agents of today work. They create templatized systems so that they just have to replace the logo as well as the name right here and then they can send a personalized package to any outreach client. Then we have the creative assets which is a zip file containing everything. These are the images created using Hixfield as well as the inbuilt images created inside Globot. Then we have the brand bible and some brand assets which are the logos, the images it has picked, the brand board which he had created earlier. Everything is inside this Google drive. Now let's check out the agent it has created. So I'll head over to 11 labs. Now inside the agent we have the healthy mid high WhatsApp support agent. It has the role, the tone, the knowledge rules. So these are the conditions for the agent to escalate to a human which it cannot handle by itself. It has also added the description of the ingredients and the price list. And let's head over to the knowledge base. And we can see that it has added a healthy mitai support. So this document contains all the knowledge base for the healthy mitai company and this has been indexed on the rack. So everything for the 11labs site and we can see the support document it has created for 11laps agent. Then since we have to add our WhatsApp number that we have to do manually because it is not able to do that by itself for now. So let's head over to WhatsApp, add an account. Let's add this one and reassign. Now once that is done, our account which is from Cutting Edge Consultants has been added to this WhatsApp agent. Now let me head over to my phone and let's try this agent out. So when I head over to my WhatsApp, let's type hi. And as I can see, the agent is replying to me in real time. So, do you see this? No delays, nothing. I don't have to employ an human to handle all the simple FAQs. I can just create a list of FAQs as well as the brand guidelines. And this agent can handle all the minimal questions by its own. Only the questions that are related to health or it can't handle will be escalated to me, the human. So, it says, "Hi there, welcome to Healthy Mai. How can I help you today? I want to order 50 packs of your famous baklava for my team. So now I've mentioned my requirement to it and let's see. So it says this sounds like a lovely gesture for your team. For bulk orders or corporate orders, please directly approach our team at customer care at healthy mitai company. This is working as I intended it to. So let's try asking a health question. Can the people from my team who suffer from diabetes have it? Let's see how it tackles this one. It says our mitai has no added sugar and is sweetening the stevia. The brand decides with diabetic friendly. That said, for anyone with a health condition like diabetes, I suggest they check with their doctor first. So, it is handling the health claims as intended by the brand. And that is how you create a very safe customer support agent which has all these guardrails around it. Now if you want you can also create your own manual guardrails. For that you have to go to this tab called guard inside 11 laps and you can add these as you want. The custom ones are the instructions that you want to add manually. But we don't require them right now because this was working as we have planned. So let's get ahead. So this was WhatsApp. But what about email? So as it turns out in the latest update by Grogbot the Grogbots itself can have their own emails. Yes, they can have their own email which they can use to subscribe to your competitor's newsletters. They can even handle objections or messages that you get on email directed to them and they can even trigger routines or automations directly from your email. So let's see how you get that. I'll head over to the chief and ask it. its email. So it says our email id is ces@mail.grogbot.com and this is because we have already claimed it. Now whenever you create a Grogbot account you will also have to create one for your Grog bots. Now mind you this can only be done once. It can't be deleted. It can't be renamed. Please make sure that you thoroughly think before you create an email inside Crobots. Now I will ask my chief to update my research agent and to its job description add a new thing. The research agent should now subscribe to my competitor's newsletters or any newsletters working in this space as well as go through the X post or the latest post they have posted on X because it can do it for free in build. So go through them and tell me what they have done till now or are there any campaigns they running. So let's get that done. Hi chief, I want you to update the research board's job description. Now, I want it to subscribe to my competitor's newsletters using this email ID, the ces@mail.drogard.com, and tell me whatever they're working on, any updates they send on mail, as well as go through their ex accounts and tell me whatever their latest updates have been, what are their email campaigns, what are their marketing campaigns, and what are the things that we can learn from them. And you should also subscribe to any newsletters working in this space, the space in which the healthy met company works in. Updating the research board. Subscribe to your competitor's newsletters at cesmail.dropbox.com. Track their email and marketing campaigns. Pull their latest X posts and also subscribe to newsletters in the healthy matai space. >> Make sure the email that you have put is right because I think you have pronounced grogbot.com but it is actually grogbboard.com. fixing that to cesmail.grockbot.com. >> Yes, please ask the research to first find out about who the competitors are and then update that as well as their emails and the newsletters they run. >> Updating the research bot's job. So, it starts with identifying the competitors then fills in their emails and newsletters. >> Also, while you're working, I want you to create an Excel tracker in the same drive. Now what this tracker is for is whatever the work that we're doing, whatever our deliverables are, they keep getting logged as per the time as well as the bot that is working on it. So that whenever I have a meeting with my client, I can just let them know that whatever we have worked on, what is the current progress and what are the things that we can deliver right now. Building that Excel tracker in the same drive, logging each deliverable with its time stamp and which bot produced it ready for your client meetings. That tracker job also just reported in. It's going into the healthy mythy drive folder and it's being filled with what's been done so far. Want me to keep going with where we left off on the research bots's first run? >> Yeah, please do. And make sure the research bot does this work end to end and submit me a list of the competitors, their newsletters, as well as the newsletters you have signed up to. >> Taking the research bot through the full loop. Competitors, their newsletters, and the signup list. Quick aside on the tracker. It's named healthy myth high work tracker in that same folder with columns for deliverable workstream bot status time logged where to find it and notes. All 16 pieces of work so far are already in it from the brand bible to the voice agent. Back to the research bot. It's running now and will bring you that competitor and newsletter list when it's done. That's the start confirmation. No check-ins. It'll click the confirmation emails itself and the final list comes to you in chat. Anything you want adjusted on that while it runs? >> No, nothing else needed it. I think the research bot has asked me a question. >> Yeah, I've approved it. >> Good. That unblocks it. Do you remember what it asked? >> No, I've taken care of it. While the researcher is on its job, I want you to add a task to its list. I Every D2C brand works with a lot of influencers. So I want you to figure out who would be the best influencers and score them or rank them in order for healthy mai company to approach as well as their contact details, their point of contacts and a dedicated brochure or an outreach deck for them which we can send directly and you have already shortlisted the B2B companies that we can outreach to using Apollo. So then go ahead and create the dedicated slide decks or the Google decks for them one by one and draft custom emails, the personalized emails to all of them inside my Gmail. >> Adding the influencer ranking with contacts and a ready to send outreach deck. >> And what about the B2B outreach? >> The B2B side is in that same batch. Decks for each shortlisted company one by one plus personalized email drafts sitting in your Gmail. Do you think there's something else that we should add to it to make it more personalized? >> A few things would sharpen it. Pulling each B2B company's recent launches or campaigns and referencing one specific thing in the email beats a generic opener every time. Same for influencers. Note which healthy brands they've already posted for so the pitch isn't cold. And if you have any customer photos or a quick behind-the-scenes clip from Healthy Mathi's Kitchen that humanizes both the deck and the mail more than any formatted slide. want me to add any of those? >> I don't have any images and yeah, I think I cut you off on the quick update. Please let me know. >> The quick update, all three landed. >> That's awesome. What's the update now? What's the progress status? >> The personalization ad is confirmed. Each B2B draft names a specific recent launch or campaign. Each influencer pitch calls out brands they've already posted for, and images are out. As for overall progress, the competitor run is still first up before any of the new work starts. and I'll flag it the moment that finishes. >> Awesome. So, while the research work is on it, I have another task for the builder bot. So, I know that you have routines built in inside your app. I want you to make an automation such that whenever a person fills an inquiry form on our website, that send button triggers a web hook. That web hook will be connected to you and with that you will take the details of that person and create a detailed personalized outreach brochure for them. This should be an emotional as well as appealing brochure for them as well as an invoice if they have shared any requirements on the spot. If they haven't just an emotional message as well as the brochure >> setting up the builderbot for that form to brochure automation. So what I've done right now is that whenever someone lands on a website and fills the inquiry form on lovable that will trigger a web hook. That web hook will be directly connected to my Grogbot routines. Now a routine will get set up such that whenever the web hook triggers the details of that person who has filled the form will come in. Grogbot will take those details and create a personalized brochure for them and put it in our Gmail drafts. So that is how you create the whole automation without writing a single line of code. Now while this is being set up, I had set something similar up earlier and I'll just show you the output of how this looks. Now I'll just head back to the lovable website that we have created. Inside this website, we have this inquiry form. So whenever you fill this inquiry form, what it does is that it simply goes to Gmail. So it takes those details that I have just filled inside the form and then automatically creates a brochure as well as a personalized message for me. It says dear part thank you for thinking us for karna consulting Diwali gifting. This is something special about a client and something related to it. It has given me a breakdown of my requirements. It says that 25 boxes into 700 rupees as approximately this much amount. And finally it has attached this brochure. Let's open and check how the output looks like. Let's download it and open it. Now it has designed a really beautiful brochure. On the header it is the healthy company's logo says happy Diwali the guily traditional sweets. A personalized thank you note. A diwali thank you from ka consulting prepared for par kana which is me. So dear part some of the warmest moments. It's a nice sweet message. The gift that I just selected, which was the sorted baklava box. And on the second page, we have the Diwali order at a glance written in a very stylized way, just like the actual brand would. And it has also added these beautiful emotional images taken from the real brand's products, but created in a stylized manner from Hicksfield. So this automation in itself as a deliverable contains a lot of value for any brand. And if you just go to a brand and say that whenever someone comes to your website instantly they'll have this personalized invoice, this personalized brochure, they'll be amazed by it. Now I'll also show you the invoice it has created. Now based on the requirement of the 25 boxes, it has taken the price of each multiplied by it and created a formal professional invoice for me. It has added an early device discount if I order before 20th October. So it automatically creates that for more offer for me. Now an important thing for you to note is that setting up this automation requires your web hook URL as well as your web hook key. And the beauty of Grogbot by design is that it is not able to copy your web hook URL or the key and paste it like it would for other tasks it has been given to. And even if I ask it to use its own computer for this task, it won't be able to do it because the guard rails around it are so strong that it simply cannot. So my step in the whole workflow would be to take the automation that has already been set up by Grogar but handling the credentials. So if you are someone who aims to be chief air officer for any company that you want to work with even if you are a freelancer who is working for a client this is where you need to act very carefully because this is the part you'll be held responsible for. So when I click on webbook URL this tab opens on the bottom we have the post the key the header. So I have to copy the post as well as the key and paste it to the agent which is making our website which is HM website. So let's copy them and paste them in this chat. Let's go to the agent working on the website. And now it'll start working on it and you have to repeat the same step on lovable as well because even lovable has this guardrail. And since the grobot you know just copy this from our chat and paste it there. This again is a step for you to take. So right now it has given me the creator collaboration deck. Let's check it out. It's a very simple and minimal deck, not with a lot of visuals. It has a lot of text which I don't think is very appealing. So I'll just head back to Grogboard and tell it. Hi chief. I just went through the output the research agent had made and it had created a deck somehow but it is not the research agent's task to create the deck. Plus the deck was really bad. I want you to be using the Hexfield assets that you just created which are beautiful assets and use those inside the decks. It should be a very appealing deck not just a very plain or simple deck. The research agents deck is out of scope and the look needs a proper overhaul. >> Correct. So now I have given my feedback and similar to the hundreds of times I given it feedback, now it will store that feedback in its memory and it'll remember it the next time it tackles something like this. And as we expected, we need to add the web hook tokens and the keys directly to lovable because grogot can't do that. So let's do it. And as a non-technical person, if you get stuck, you can just simply ask robot, please tell me where I can find the web hook token. So, this is my web hook token. I'll just copy it, head over to Lovable, and paste it. Now, as I mentioned earlier that it was not the research bot's job to create the deck in the first place. The chief has realized that and then delegated the same job to the creative bot. Now it is using hex field to create that and adding all the visual assets to the Google slides deck. So let's see the output now. So the research agent has given me an update. It has created those 35 Gmail drafts. Why 35? because it was not able to find uh the email ids for 18 companies. So it has not assumed them which is a good thing and while that is happening the creative agent is working on the decks. So one agent is working on the research one on the decks and everything is coordinated by the chief of staff while in the back end it has the memory as well as the brand bible. This is literally like an army working for you. and arm your bots working towards one end goal which is to cater to your client. Now the inbound is handled by the automation that we have set up. So till now we have handled the brand assets, the image ads, the video ads, the inbound, the outbound outreach. Now we just have to wait for the decks and these will be drafted in our Gmail drafts. everything attached like the brochures that I shared with you and everything will be ready for us to send. Now that you have seen these possibilities, there are actually two ways in which you can sell these assets, the deliverables to your client. Either you can sell the outputs and charge for your judgment by picking the right outputs or you can set up these agents for your client. Now, how do you do that? So just like we had seen the marketplace where all the Grog bots were listed, you can also create your own Grogbot template which you can just send to your client. So let me teach you how to do that. Let's go to any what let's say research. On the top right you have this icon. You have create a copy. Once you click on that, there will be a message triggered which will say create a public template of yourself that I can share with somebody else. In a few minutes, everything that it has learned so far, everything that we have taught it, everything that Chief has taught it will be in its memory as well as the skills. Now let's see this card which says research. So it has all the context, it has the memories, it has the skills. Everything that we have worked on till now has been converted into these three which are the instructions, memories and skills and they're ready to be published. So if I just click on publish, I'll get a sharable link. This link can be shared with my client. Now imagine this. I can simply go to the healthy mai company and offer them a roster of these agents. I can just send these to them so that not just Diwali but for any festival any event they can use my bots and I can charge for creating these agents. As I come back to my chief it says that the B2B master deck is ready and the creator outreach deck is also ready. Let's take a look at it. And this is the output it has produced now. Can you see the difference? Earlier it was such a simple plain deck and right now it has the branding, the images, the typography, everything and this is because the earlier deck was created by the research bot but this was created by the creative bot. The creative bot is a specialist in this. It has been trained on this only and that is why you should have one bot one job. You can see the output in front of you. So we have all these slides, the collections, the quality, the timelines, delivery and refunds and the contact details in the end. Just look at the quality of the output. And if you open the creator outreach, let's look at this as well. So, so this is the creator outreach deck. Again the logo is consistent, the images, the products, the content ideas. So it has gone ahead and created five content ideas, put it in the creator deck along with the images, everything is done right here. And that's amazing because inbound is handled, outbound is handled, customer support, assets, everything is handled end to end. And we haven't drawn a single image. We haven't written a single line of code. Everything was handled by Grog Bots. The place we came in was a judgment. We told them this is not how they should work. It should be done by this one and like this. And this is the output that we have received now. So let's go ahead and praise the chief and tell it to now create the outreach emails. So I have asked it to create five personalized emails for B2B brands and five for the top content creators and right now it is writing the personalized emails as well as attaching their personalized decks. So now the whole army is working for you and you also know how to package and deliver this to your client either the assets, the automations or the agents. themselves. And now comes the final part. How do you find clients for yourself? So let's just head over to the chief again and let's ask it. Hi chief, while you're at it, I think we have done most of the work that we're intending to do for the healthy mai company. We'd like to do the same for the next 10 D2C brands. Please create a short list of them and give me their PC email ids from Apollo so that I can send an outreach email personally to them. Short listing 10 DTOC brands next and pulling their PC emails from Apollo. >> Thank you so much. So while that is happening, we have created this army for our client. We have made the deliverables. We have packaged it. And now I can repeat the same process for the next 10 D2C clients. And this is how I as a oneperson company can scale. Now mind you this is not about replacing people. This is a force multiplier. So if you have a team already they can come on the team plan of the Grog bots. They can assign tasks to each other and they can use their expertise on the judgment rather than focusing on the grant work and everything it needs to save it comes with an approval card. So yeah, always allow. So it has gone ahead and shortlisted these 10 brands for me to work with next. So now as I head to my Gmail, I can see the seven drafts ready. I can see to Lehi and they have written a personalized message. It has somehow added a few links. I don't know for what reason but this is not how a mail should look because the message is good but you shouldn't be reiterating the links and that too long links. So now I'll just ask my Grog Bots to never do this again and they'll remember it. Hi chief, I came across the mail that you have written. The drafts are containing links for some reason. They shouldn't be there. These should be mail that are crisp, emotional, and to the point for the person to just read and connect with us. >> Stripping those links out and tightening every draft in devicemail.com into short, emotional, no link versions. That's the rewrite running now. Want them to open with a personal hook about each recipient or keep it more about what device MC does? >> Yeah, correct. Keep the hook personalized to them what their recent updates have been. But remember these things for the next time. Add this to your memory. >> Personal hooks from their recent updates noted for every future draft. >> Awesome. So just like that again something I didn't like, I told it. It took the feedback updated its memory and the next time it repeats the same process, it will not make the mistake. And that is how you improve your grog balls over time. These aren't perfect out of the box. That is for certain. This is like a new hire. You keep them with you. You make them see your work. You train them and then you teach them the tricks of the trade. That is how you make an army of bots to work for you and work like you. And based on my feedback, it is updating all the drafts again. And eventually it'll add the personalized decks to each and every draft from the main master deck that we have just reviewed. So throughout this course the bots did the work. A human approved it. Now the button of this UI might change but the way of working won't. So now your homework is to download the kid document that you have attached in the description. Pick any young brand that you love and try to run some of these use cases for them. Don't email anyone. Don't outreach to anyone right now. First improve your bots, improve their outputs, verify them yourselves. And try to keep a filter that is this something that you would buy as a first-time buyer. Is this output appealing to you? Is this something that you would actually use? And once you verify that only then you should outreach to other brands and show them your work. And that's all from my side. Thank you. If you're watching this video till this point, I can guarantee that you're super motivated to at least build and grow in the age of AI. And I know that you will say that an why can't I get grock for free? Is there a free tool available? And you know that the answer is that nothing valuable in life is going to come for free. I can give you tons of videos where I will say build a oneperson business using free AI tools. It's all crap. It's all nonsense. It does not work. We've been building solutions for the last 2 three years. So I'm telling you, you have to invest some money into this to build real skills because if you genuinely want to build a business and actually be productive, you have to invest the time and money to learn a new tech in the next 2 to 3 years. People who start learning today will have a massive advantage. Job loss is on its way which means that a lot of people will be forced to actually start their own business and I'm not saying that learning gro will make you a lot of money or it will make you immune to changes because honestly everything that you learn today is going to lose value in the next 2 to 3 months. That is why we recommend people to learn more about the shape framework where a stands for arbitrage. Everything that you're learning today is going to become valueless in the next 3 months. But all of these learning sprints will help you refine your pattern finding. It will help you understand how the world is evolving. And these sprints will help you enter the game early and stay more aware than the others. The gap between the rich and poor is going to become super massive. And my hope for your career is that you are in the rich category and not in the broke category because awareness compounds over time. So if you start learning today, you will pick up so many nuances and little little details over the next 3 years. In the next four years, you'll be the only person with four years of experience deploying AI agents for an enterprise. And that will count a lot. Now, if you want guidance on how you can sell these agents as a service, we've already created a two-part series on how to start your own agency. It has extremely practical advice on how to start and how to price your work and how to find clients. The link to those agency videos is in the description. Otherwise, just go to my YouTube channel, type the word clients or type the word agency. You will find everything. Do not forget to click on the subscribe button so that you can learn from the upcoming episodes as well. If you have any doubts or any suggestions, please feel free to ask in the comment section. Our team will help you with them. If you document your notes in the comment section of this video without using AI, you will stand a chance to win a 24,000 rupees scholarship from the Cutting Edge Academy. Details are in the description. Take care of your mind and body. This is your host Anmehra and you are learning from the cutting edge school.
Assistir ao vídeo original no YouTube
Assuntos: Agentes de IA, Assistente pessoal. Publicado no YouTube em 2026-10-20.
Seleção automática por assunto e idioma. O vídeo pertence ao canal de origem.