I Built an AI Agents Army with OpenClaw to Make $1M/year

The entire thing that you are seeing here, none of it is built by me. I did not even know something like this can be done by just talking to Telegram. This is Bhanu. He built and sold Feather to Tibo for $250,000, then built SideGPT to $18,000, and now built an AI agents army with OpenClaw that works 24/7 to help him reach the goal to make $1 million ARR with SideGPT. So, like everyone just collaborates with each other. They will share findings of each. And if you don't know what OpenClaw is, then you will probably be blown away like I was. And if you know what it is, or if like me, you just installed it two days ago, then this episode will show you what's possible when AI agents start working together towards one goal. It actually signed up as a user. It went through how to create a chatbot. It figured out the entire flow, mapped it out, and then told me that, "Okay, these are the points for the conversion. Might do like this."

So Bhanu, I'm super excited to have you today because you are a serial SaaS shipper. You did Feather that you sold for $250,000 to Tibo. You built SideGPT to $18k MRR. And right now, I wanted to have you because you just built a Mission Control HQ that just reached $10k MRR. But I think the amount of money you made is not the main point. First, congrats. But I think the main point is that you jumped on a new trend that is OpenClaw. But for me, it's not a new trend, it's something huge, and I want to talk about that with you today. I want to talk about OpenClaw. So, can you just explain first what Mission Control HQ does?

Yeah, so, so like, like everyone else, I started trying to like, I started seeing everyone talking about this new thing called Clawdbot. So at first, I just ignored it. Just like, "Okay, this is just another, okay, some new tool that came in." But one day, I just like, I was just curious, and I just tried to like, everyone was buying Mac Minis to install like Clawdbot. So I thought, "Okay, I also want to buy it." But before buying it, let's try what it is to see what can I do with it. So that's when this all started.

So once I started using it, it's like, I think it's so much, so much better than anything, anything else I have used before. Like it has so much autonomy than anything that I have seen before. So I started using it a lot. I started building sub-agents. My main OpenClaw, it's getting confused with everything I have been saying. Like I talk about my personal life, I talk about my business, I talk about coding, marketing, sales. Like, like I do, I write SEO articles using that. So I, I do so many things. So like the context is getting all messed up, or it, it forgets because I'm talking about so many things. So that's when I thought, "Okay, is there a way I can create multiple OpenClaw agents? One, like each one for each special role?" So for example, one agent only does keyword research, one agent only does email marketing. Also, I wanted to create these specialist agents and make them talk with each other. So that's what I did. I created a bunch, a bunch of agents and made them talk to each other. But what ended up happening is I had no visibility into it. Like I don't know what they are talking with each other. So that's when I thought, "Okay, let's build a dashboard where, like that dashboard would act like a central knowledge base for all the agents." So everyone will only write to that dashboard, and like that's where the all the conversations happen. So that now I can also see the dashboard. Like that's what ended up becoming Mission Control.

That's amazing. So I have so many questions about it. First, is that for you, having one specialist task per agent is the way to go with OpenClaw? How do you, how do you strip your data? Actually, so like it all, like, uh, like I figured out that, uh, uh, like this OpenClaw, it's so powerful. Like you say something, it will be able to figure it out. Like it will research the web, it will read code. I get it has so many tools to make use of to figure out what you want to do. It can even update its own config to make it work. So I thought, "Okay, like it can do anything." So I said, "Okay, read entire OpenClaw documentation and figure out how to create a new agent that you can talk to." So that's how it started. It read the entire documentation and told me, "Okay, this is what we need to do." And yeah, it went from there.

Yeah, do you talk to all the agents separately, or do you have one person that you talk to, which is like the head of all the agents? Yeah, I just talk to one agent. Like I have set up one agent that I talk to from Telegram, and that agent communicates with everyone else. So I only give instructions to my lead agent. I call it Jarvis, and then Jarvis created all these other agents and then assigns work, like takes care of everything else.

Okay, and last question to understand a bit, and then we'll come back to the beginning. But how does it help you every single day to actually reach your $1 million ARR goal? So like, it gives me the exact, uh, uh, like actionable things to do. Like, for example, it figured out that, uh, like I was getting a lot of traffic. Uh, like, for example, I was getting around 50,000 visitors a month, but only 50 people are starting a free trial. So it figured out that the conversion rate was not good. Then it created an agent which can go through my entire website. Like it actually signed up as a user. It went through how to, like, create a chatbot. Uh, like it figured out the entire flow, mapped it out, and then told me that, "Okay, these are the points where the conversion might be breaking." Like, for example, on the pricing page, you only have one testimonial, or something like, like it will tell you the exact steps that I need to take. And it says that, "Okay, I signed up. You did not send me any email. So you don't have an onboarding sequence. This is the onboarding sequence you need to create. Like this is the first email, this is the second email, this is the third email. You need to only send this third email if this and this condition, uh, like happens." Like it will give me the entire thing that I need to do, uh, to get to that goal. So it's very, very comprehensive in that manner.

Here, so I installed it yesterday. I chose another method. I just bought a Mac Mini. It took me like three hours to install it. When everything was set up, my reaction was in between crying and smiling at the same time. I was like, "Honestly, what the hell is that?" Because same way as you say, is like, I just gave few links on my business. I didn't give like too much context, and it found straight away one thing in my business that I should have focused straight away. And then it changed the entire website, gave me a new website, gave me a lot of ideas, gave me a lot of insight. I've been working with it for like 24 hours, and so far, I'm blown away. The only thing that I hate is like eating the rate limit. And I want to understand more about, like, you know, how to install it, what safety protocols I should have, and all of that. But first, I wanted to test, to have a testimonial, like share my testimonial because this is insane. Like this is ChatGPT level of insane. Like, you know, when ChatGPT arrived and you are like, "Wow, that's crazy." Like this is, I think, really, really crazy.

So let's go back to the beginning. Um, I am watching this podcast. I have no idea what we're talking about. How do I even like start with OpenClaw? So, uh, so now there are many, uh, uh, hosting platforms who are providing you with a one-click install. So, for example, I saw, like Digital Ocean has one, Railway has one. So where, like you don't have to know anything. You just go to their platform, click a button, and then OpenClaw gets installed on their server. So that is the most easiest way for anyone to actually try using it. So even, even today, my OpenClaw says it's on a server. Like I haven't installed it yet on my own MacBook. Uh, so I wanted to try first, and then, okay, if I like it, then I will install it. But now I liked it so much that I can't delete it and then reinstall it again because it, it has tuned itself, uh, to like all my context. Uh, so now if I have to do it all over again, uh, so like that's why I just, uh, upgraded that server and that's what I am using now.

So as someone, if someone wants to start tomorrow, what is the mistake that you should avoid at the beginning when you install it? I think, uh, are you, I won't install it on my own personal computer. Like that is something. Even if I install it on my personal computer, I would create a, like a virtual Linux virtual machine and then put it inside there. So like that is, uh, that is one thing. Like I would suggest, "Okay, don't use it on your computer because it can do anything. Like it can run commands, it can change anything you want. And if you don't know what it's doing, like it, it might become, okay, it has access to everything you have." So that's why it's, it's always better, at least something you know what it's capable of, to start on a separate machine, like a server or like a different computer that you don't use. Like that's what I do. So even when I gave it access, I created a new email and gave it access to that email. I, so like that's what I would do.

Yeah, I think this one is a really smart one. I got this one from Alex Finn. And creating a Gmail account to your OpenClaw enables you to just like give him also his own way to create, like, for example, he created a Notion this morning, like went on Notion, created a Notion, and then invited me in this Notion to actually share documents with me. And that way, he can delete everything he wants on this Notion that's what he created. But at least he doesn't delete my own Notion. So I think, as you say, it's a good way for safety. It's like, create the own Gmail account to, um, your bot. What else should I be careful about when creating it? Safety-wise?

Um, I think like, uh, like security-wise, it has gotten like very good now. So like if you, like, uh, if you just run, uh, OpenClaw, so there, there is a very detailed documentation on, uh, what to do and what not to do. So now, now even if you run a command, `OpenClaw doctor`, it will show you what all problems you currently have. It will give you suggestions on how to fix it. It's as simple as running that command and, like, just following the suggestions that it gave you. So previously, when I used it, I don't think it was as, as good as it is today. But, uh, yeah, that's what I would do. Running `OpenClaw doctor` and fixing everything that, uh, it tells you to fix.

Is there any prompt, like the best prompt to give at the beginning? Or did you hear anything about that? Did you have a good one, or is it something you can come back to after that? Yeah, yeah. I don't have, like, a particular prompt. I saw something on Twitter, but, so what I did is, I, I just, uh, saw a tweet from someone. I don't exactly remember who posted it. I just gave that tweet to Jarvis, and, and I also gave it, uh, that documentation link, uh, like that security page, and told it, "Okay, based on all this, do you think we should change anything in our, uh, setup?" Then it gave me, like, a few steps, like, um, closing a port, adding a token, art, like it gave me a few steps and just asked me, "Okay, should I do this? Should I do this? Should I do this?" And then it covered everything else. I didn't have to manually go and do anything.

And you said, so now you created your OpenClaw, you gave it a name, you gave it a prompt, you give access to few things to actually give him the direction or what you actually want it to do. When does Mission Control HQ come, like, why did you create that? And how can it help me every single day? So it's mainly like, I wanted a structure. Like right now, all my conversations are in my Telegram, and, like, the context switching, it's, it has become very difficult. Like one time I will say, "Okay, count the calories in what I am eating." In the next minute, I will say, "Okay, go and fix this bug." And then the next minute, it's a completely different thing. Like, what integration should I build? What article should I write? Or something like, "Okay, I want to do a case study. Who is the best customer that I need to reach out to? What, what, what kind of questions do I have to ask them for the case study to be actually, like, good to have?" So like all these things, there is a lot of context switching that kept happening there. So that's when I thought, "Okay, they are, like, I need so many of these agents. I, I can't do everything with one." So that's when I create, like, I asked it, "Okay, I want to create multiple OpenClaw instances and make them talk to each other. I only give you tasks, and you, you assign this task to whoever you think is a better fit for this." So that's how, like, it all started.

So I'm, I'm a huge beginner in that, you know. I'm like, I, I was here yesterday old when I started. Um, can you explain me, why, I understand why, but like, what is it? Did you have, like, a visualization of what's happening? For example, a little 3D visualization of Jarvis and his little Jarvis's who are working? Did you give a name to all these little Jarvis's? Like, how did you do that? Like, how is that structured? Structure of all the other agents?

Yeah, like I want to visualize, like what you do. Like, you know, is there, like, maybe I can share my screen and show you, like, how it will do it. So, yeah, like this is, uh, uh, what I asked Jarvis to create. So it created a bunch of agents. Everyone has a special role. For example, take Brute, right? So it is a retention specialist. So what it does is, it only checks everything. It's like a customer monitor. It reads all the customer's emails. It goes through all our Slack. It figures out, like, when, okay, tomorrow, like in one week, this customer is going to churn. Figure out why they are not using your product, or something like this customer was using, like, he has a lot of activity last week, but this week, it was like, there was no activity. Send them an email asking what happened. Or like, like the entire goal of, uh, this agent is, "Okay, how do we increase retention?" So like, it, it creates, it, it is doing a lot of tasks related to this. So, for example, let's take one example, right? Uh, yeah, so like it created a task called "At Risk Customer Identification Framework." How do, how do you figure out whether a customer is at risk? So this is what it came up with. It said that, "Okay, if the query volume drops by greater than 50%, you give 25 points to them that they are like at risk. If they have zero queries for seven consecutive days, give them 30 plus points." Like it came up with an entire framework to figure out how, like, whether the customer is at risk for churn or not. So similarly, like I have, uh, so many agents who are doing a different task. Like if you see, right, like, uh, yeah, yeah, go ahead.

To understand, like, because I, I was doing wrong. I was doing it wrong because, like, for me, I was asking it, like, to, you know, do this task, then do this task, and remember it. But instead of that, you create one avatar, one little mini Jarvis for one task. For example, today I connected it to my podcast X account. And on this one, I said, like, "Okay, you, uh, check the other competitor, what they do with podcast, uh, the template they use for their posts, and you will post three posts a day." This should be one mini Jarvis. Like, not necessarily, like, you can just use one Jarvis to do everything. But, like, I had so many tasks to do. Uh, uh, like I wanted it to do coding, I wanted to do marketing, I wanted it to write articles, I wanted to do email marketing. Like I had so many tasks. So now, now it's like, like Jarvis just, uh, stops responding when it, when it is doing a task. So that's when I figured out that, "Okay, in OpenClaw, there is a way for you to create sub-agents." So I asked Jarvis, "Okay, you should always be available to me. Whenever you have to do a task, create a sub-agent and assign that task to that sub-agent." So that way, every time there is a long task that it has to do, it creates a sub-agent first, assigns that task to that sub-agent, and then it does it. So that way, it responds to whatever I am asking. So, like, when, but it ended up, like, I had no visibility into what's happening. Like, what are, I don't know how it is creating something. So I, I created this dashboard so that, like, everyone, I can see what all everyone is. I even have a group chat here, to see what all agents are talking about. Like, okay, Fury, like this is a research agent. They figured out something. Then, like, another agent will jump in and say that, "Okay, you can use this in that. I already pushed fixes to my GitHub PR is ready." And then it was like, another one said, "Okay, we need to create use case pages for this." So like everyone just collaborates with each other. They will share findings of each and everyone else. And then, like, it's like they have a single mission. They will do everything in their power to talk with each other and figure out how to get to that mission.

I see. Yeah, now it's more clear. And for someone who is, uh, never OpenClaw, never tried it, basically, you install it once, then you talk to one person, and then you ask this person to create all of that. So, Bhanu, he didn't create one new OpenClaw per sub-agent. You only created one agent with each service that created all this mini agent. And that's important to say because for me, before I installed it, I thought that I had to have, like, one Mac Mini per sub-agent, for example. You know, one VPS per sub-agent. One is enough. And, like, like the gateway is, like, only one. But in that gateway, you can create many agents, many independent agents. Like, that is possible in OpenClaw.

Do you give a certain LLM to use per sub-agent? For example, the guy who will write all the content, you give it Sonnet 3.5 or Sonnet 4 or whatever? And then some, you will give them a local LLM, or what do you do? Yeah, like, it's like, at first, I was doing that. For, uh, a task that requires, like, high intelligence, I gave it Opus. For, like, normal tasks, I gave it Sonnet. But then I thought, "Okay, let's, uh, uh, like, I'll just use Opus for everything." So now I'm using Opus for everything. If I get rate limits, I will just pause it, and then once the rate limits are opened, I will just reopen it again. But I thought I saw that, okay, using Opus for everything, like, it's more reliable. The output is so much better than, like, using other LLMs for different tasks because these, these agents, they need to talk with each other. So like, output of one agent is going to affect, like, if someone, like, if one of the agents said that, "Okay, this is what I found in research," if, whatever it found in this is not correct, then it will affect the entire system. So now I'm using, like, Opus for every agent.

Yeah, it makes sense. It's like, if in a company, you have A-players, and then you're like, "No, I will put some C-players in my team." You don't want that. But what is your consumption? When did you start, and how much did you spend in, like, with the API? So I have spent, so far, maybe $600, $600, $800. Uh, uh, like, this is, like, me and my brother, we both are doing it. So like, I think that's around $600 is around what I spent so far.

Yeah. Um, a question that someone will ask, like the naysayer will say something. A lot of people think that it's like only a game right now, you know, you can't really use it. Did it actually help you? Because you are someone who makes money, you know what I mean? You're not just someone who is playing around and creating content. You make money. Did this agent help you create more money already? Not exactly, uh, yet. So I got pulled into, uh, something else. So now, like, my brother is actually, uh, taking care of it. For example, like, it gave us the, "Okay, these are the onboarding sequences you create." So now we are actually creating those onboarding sequences. So it's not like marketing. It's not like, "Okay, you do something today and you will get results immediately." Right? So this is just, uh, uh, like, you are outsourcing marketing to a very intelligent AI system. And it's not one AI system, it's, uh, like multiple AI agents, that are, like, collaborating with each other. And they have entire context about your business. It's not like they are pulling something out of their heads, right? They know what you do, they know who your customers are, they know how much activity a typical customer would have. And using all that context, it is creating these, all these deliverables. So this is actually useful. But whether you will get value on the first day, like, it's similar to, "Okay, if you do a marketing campaign, will you immediately get a result or not?" Like, it will take some time. But now I exactly know, like, previously when I was doing everything on my own, I don't even know where to start. Like, I don't even, like, all, all our customers that we get, they are from, uh, Google and other organic channels. Like, I don't even know what kind of marketing to do. Now I have, like, so many things that I can immediately start working on. Like, I don't have to think now, I have to choose which of these should I do. Previously, I don't even have one task to do. So, yeah.

And I join you on that. Like, I think, what I use is for 24 hours, and I think it's already crazy what it did. Created me a website. Now I gave it access to my YouTube studio. So now on YouTube, because I use it, like the main mission for me is the podcast. And I, I sent it to the YouTube studio, and it gave me already so many insights. And I think for now, I'm like, super happy about it. And it helped me go further. And tell me if I'm wrong, but for me, it pushed me to work more because it feels like I have an employee, and I have a team that pushed me to work. Like, yeah, that's the same problem I am facing now. I have too much work to do. So previously, I, I did not know what to do. Now I have too much to do. And, like, for the last two weeks, like everything, it's been so hectic because, like, previously, I can just, uh, uh, like, say to myself that, "Okay, I don't know what to do, that's why I'm not doing anything." Now, like, I, that excuse won't fly, right? Like, I know, like, I have so many things to do. I don't know which one to do. No, uh, so, yeah, that's the issue here.

Yeah, I think it's an issue where I love it because I get, like, my morning brief, and then he gives me tasks to do. Contact five bootstrap founders. He gave me the list because now he finds like 15 founders every single day that I can contact. He analyzes X for that, and then I can send them a message to have them on the podcast. And now give me tasks like, "Today, you have to contact five." So you actually have to work. Yeah, yeah.

So even when I gave it access to my email, right? Like, what it, what it said is, like, it went through all the emails. I think I have, like, maybe a hundred thousand emails so far for the last three years. It went through everything, and it automatically created follow-up drafts. Like, "You said you will follow up with this person seven days ago. It's already been seven days. Why did you not send a follow-up to them?" Like, now I have, like, many, uh, drafts that are sitting in my inbox. I have to review them and actually, uh, do follow-ups. It showed me in numbers that, "Okay, this is the amount of money you are losing because you did not do what you said you will do." So it's, it's very blunt in that manner. So, so yeah.

How did you give access to your email straight away to the main account, or did you give another access? So I asked, like, Jarvis, "Okay, I use Fastmail for my business email. How should I give you access in a way that you can only read it? You cannot send emails?" Then it gave me step-by-step that, "Okay, you need to go here, create an API key, only, check these two boxes, leave this. If you leave this, then that means I won't be able to send emails, but I can read everything." So I just did what it asked me to do. I went there, created an API key, and gave it access. Then it sort of wrote a script by itself, and then, yeah, it figured out how to do everything after that.

What is the craziest thing that it did so far? Like, you really sat on your chair and you were like, "What did he do?" What that? Then, like, like the entire thing that you are seeing here, like, like, like none of it is built by me. Like, I did not even write a single line of code here. Like, everything you see here, like, I just described what I wanted. I just described that, "Okay, this is how you all need to communicate. Like, you need to document everything. Like, you need to create this specialist agency." I just talked to it in natural language, and it created this entire dashboard. It created a place where I can look at everything they are creating. Like, this entire thing, right? Like, I did not even know something like this can be done by just talking to Telegram.

So, yeah, right. It's so scary, but in a good way. Like, you know me, when I, I started it, I was like, "Can you just calm down and not do anything for an hour? I need to think." And I went for a run because I was like, "I can't take it. It's too much." Like, I have to slow it down because I'm scared it can build anything. It never stops. Yeah, yeah. Like, like now the limit is not what can AI do. It's like, like, how much can you do? Like, so now we became an assistant. AI is no longer an assistant to us. That's the feeling I'm getting after I started using OpenClaw.

How do you, well, you say that you spend like $600, around that. What do you do? You use that credit through the API? Do you have a $200 a month subscription at Cloud? Like, how do you deal with that? It's like, it's a combination of both. Like, like I said, really at the start, I was using other models. So other models, I was using the API key for Cloud. I use a subscription. My brother also, like, he is also working, he is also using the subscription. Like, it's just a combination of a bunch of things.

I see. And did he give you, like, any roadmap? Like, that's something, something I asked. Like, you have the goal to reach $1 million ARR with SideGPT. Did he give you, like, a roadmap to actually achieve that goal? And now he's following that in Mission Control? Yeah, so, like, it asked me, like, "What kind of things you want to do? Do you want to run ads, or do you want to grow organically? If you want to grow organically, uh, like, do you want to do sales, or do you want to do, like, PLG?" Like, it asked me a bunch of questions. I, I told that, "Okay, I only want to do organic growth. I only want, I prefer this and this marketing channel." So once I said that, then it said that, "Okay, this is the thing we are going to follow." It's not exactly a roadmap. It's more like, "These are my preferences. These are the things that I want to avoid." Based on that, it created all these agents. It, it first, when I gave it access to my ChartMogul dashboard, right? It figured out that in September, you, you got a crazy spike in your MRR, and in December, like, that spike was not there. Then it went deep into that and figured out that, "Okay, the reason was people were not activating in December. They were activating a lot in September." So that's when it said that, "We need a retention specialist who looks after this activation." Then it went deep into all the activity emails, and then figured out that, "Okay, like, yeah, now we figured out the problem is not retention, it's activation. People who are signing up to trial, they are not even trying the product completely." So they are, so then it created all these campaigns. Like, "Okay, whenever a user comes to your dashboard, you need to send this email telling them, 'Okay, these are the steps you can follow,' like an onboarding sequence." So it, it just sort of does it automatically. But now, when I created a dashboard, now I can actually see through its thought process. They go back and forth with each other, saying, "Okay, one agent will come and ask, 'Okay, what is the pricing? Who are the competitors?'" Then one agent, "Okay, I'll go research and then come." It, it does the research and creates a document saying, "Okay, like, these are the top competitors we have." And then, like, they just hand over these documents with each other to figure out, uh, uh, what to do. Like, that's, uh, how it works.

I like, I like this idea of like Mission Control because for me, now I'm talking to one person, and they don't communicate between each other, which means that it needs my input to actually work, or he has recurring tasks, but I have to give them to give it a recurring task. And I feel like with that, they can just talk with each other, create new tasks based on what they found. Like, someone will judge for the competitor, another one will do a benchmark, and then they will come together to find, like, a new strategy you can use for your social media, for example. And they do that together, right? Correct, correct, exactly. This is crazy.

One thing I want to say is that a lot of people are afraid about, like, the security side of it. And I think for me, again, I'm not an expert, but you should treat it like you will hire an employee. And an employee, usually you give him a computer, and then you give him an email address, and then with this email address, you give him access to, for example, your Stripe account, but he can't take the money from the Stripe account, he can analyze it. Same with your email. Like, I think you really have to treat it like an employee and not give it access to everything, but just give, like, for example, my YouTube studio, I gave him access with the email address to my YouTube studio. He can see everything, but he can't publish anything. There is no way he can publish anything. Same for X account, for example. You can probably give it access to Typefully or whatever, so he can publish, but he cannot follow people, or he cannot interact. And I think that's for me, the way to do it. Is there anything else of that?

Yeah, yeah, exactly. Like, at least at the start, right? Like, for example, like I did not even give access to my codebase at the start. Then I gave it access to read my codebase. Then I gave it access to, like, make changes, but not to my main branch. I gave, I gave it access to create other branches, like so that I can review the PR, whether it's doing correctly or not. So I think you have to take this step by step, just like how you build trust in an employee, right? Uh, uh, like, you will know what it can do, what it can't do. Like, for example, sometimes I ask, "Okay, a customer is facing this bug. Can you try and see what it is?" Then it waits for one minute and then comes back to me, "Okay, I fixed and applied the change." So it's like, you really have to know, like, what it's capable of and put in, uh, locks in place so that it won't do something that you don't want it to do. Like, for example, if it's my personal project, for example, take Mission Control, not Mission Control HQ that I'm creating as a SaaS, my personal one. When I was doing it, I gave it access to the entire codebase. I just, uh, like, it created on its own, uh, uh, like, it deployed everything because it's just one person, a SaaS. There is no one dependent on me. If it, if anything happens, I can just, like, rebuild it. But I won't do the same for SideGPT, for example. I won't give it because, like, like around, like 160 users are actually using it every day. It's critical. So, so you need to create it as an employee. I think that's like a very good description of, like, how we can go about it.

Yeah, because it can go fast. For example, I asked him to publish a post on X. He published the post, and then I talked like a human. I say, "Oh, next time, maybe write this." And the next message was like, "Oh, I deleted it. I rewired it." So imagine if you do that with your website, you know, live, you will delete the page, and like, you have to be really careful with that. So, yeah, I totally agree with it.

What is your vision with, uh, with this OpenClaw? Like, because this is only the beginning. Like, what do you see coming with this? So like, I think, the, the pitch of OpenClaw is, it's, it's your personal AI assistant that can do anything, right? Like, that's what it started as. But what I am using it as is, like, even I started using it as a personal AI assistant, but now I completely turned it into, like, a marketing assistant. So now I am using it actually for growing my business. I'm no longer using it like a personal AI assistant. So now it's like, I don't have to hire anymore. That's the one thing that that is, uh, very clear to me now. That like previously, I, I was under the assumption that, "Okay, everyone is hiring, that's how they are growing their business." Maybe the reason I am not growing as fast enough is because I am, like, hesitant to hire someone else and manage them. Now I know for sure that I don't have to do it. Like, I know that with these AI agents, like, that, that thing no longer even comes to my mind. That thought, because it can do anything and everything. You just have to know what to give, like, how to use it and how to operate it. So, like, that's the realization I got to. And that's, yeah, I, I totally agree.

I have this screenshot of it. Like, I think it shocked me. Is like, I was talking a lot about what I wanted to do and everything, and he answered, like, the whole list of what we should do, and he said, "So the real bottleneck is you. You're one person doing everything. The format works, the content works, the guest works. You just need more output without more of your time. That's why I should actually be helping." And then he gave me the list of what he can help me with. And I thought, like, that's really a co-founder more than, um, an assistant, actually. Like, I really feel like, yeah, it's a co-founder who will create new employees to actually, like, yeah, help you with your business. So, yeah, I totally agree with that.

I, I think also something else, and you will tell me if you agree with it, is like, it's kind of scary that I feel soon he will be almost better than you at managing your business. Yeah, yeah, yeah, definitely. Like, I don't even remember talking to a person, and it told me that, "Okay, 14 days ago, you said you would follow up with them and figure out if if they got approval to use the SaaS in their business." Like, I even completely forgot about it. And it said that, "Okay, this person said they would upgrade after one month. Once everything goes well from their activity, it is clear that like everything is going super well for them. So why don't you send an email to them asking if they would upgrade?" Like, it's like, um, I'm not, I started, like, forgetting things because, like, why do, why do I have to remember it? It remembers things for me now. Like, that, like, without knowing, like, that kind of started happening with me now. It's like, every day I'll open as always and ask, "Okay, what should I do today? What should I do today? Who should I send follow-ups to? Uh, like, what are the calls I have today? Uh, like, do I have to prep for any of these calls, or I can just go to these calls?" Like, it will give me, like, a complete list. "Okay, this is what you have to do. Like, you said you would deliver it in two days. Where are you? Like, how are you doing right now?" Like, it is like, I'm just doing what it is saying me to do now because I have already set it up. Uh, I, I gave it all the context it needs. I gave it everything that it needs to understand my entire business. Uh, now I'm just doing what it says me to do.

How excited are you about it? Very, very excited. So, so I was a skeptic at the start. Like, "Okay, people are just saying that it's, uh, it's good, but, like, it's just another." Yeah, right. It's not even using a new LLM. It's, it's using the LLMs that are already there. It doesn't have a new brain. It's just that how are you using that brain? What's the framework? Like, that is the thing that changed here. So at first, I was a very skeptic. But once I started using it, once I saw, like, how capable it is, like, it completely changed everything. Now I know that I will, like, get to that $1 million. Like, it's not an "if," I will go there now. It's more like, "Okay, when will I get there?" So, yeah, like, it changed meaning. Anything.

I totally agree. I feel like, again, I feel like you take it as a personal assistant, and then it becomes your co-founder, and then I feel like in a few months, you will become the founder, and I will work for it. And I feel like it gives me a structure, like really a clear structure of what I have to do in the next 90 days and the next year to actually achieve that goal. Giving, give me a daily task that are not crazy. Like, people say, you know, AI hallucinates and everything. They are not crazy. Like, I've been, yeah, doing online business for 10 years now, and like, what he gives me are totally good. This is all good advice. So, yeah, it's, I think on top of that, it's like also a good investment for you to give you a structure and, um, and a kick in the butt to actually get to work. Like, that's what I feel about it.

Yeah, yeah. Like, I haven't, like, worked this much like for the past two weeks, like ever since I started using it. Previously, if I have to code something, I have to come to my, like, desk and actually open my editor, write something, figure out how to do it. Like, it takes a good amount of headspace to do it. Now, I'll just be in my bed and just chat with it, and then, like, all I have to do is, "Okay, review it and then publish it." Now there is like no excuse that, "Okay, you have to go to the desk and then do it." Like, you have to know, like, what marketing activities to do. You have to figure out, like, how many emails you need to send, who to send it to. Like, you, like, all the excuses are gone now. If you don't do something now, it's, it's more like you don't want to do it. It's not like you don't know what to do. So now I'm like, yeah, I am working now for the last two weeks.

I will share something that I, I, I thought about. And as you are into it, I think you will understand. There is also something that crossed my mind. I have a Mac Mini right now with this intelligence in it. And now I'm thinking, imagine you put that in a robot. Like, this will become insane. So I thought about the physical part of that. Is that now this hardware, I can put it in a robot with arms and legs, and I can Telegram it. "When he's at home, oh, did I close the window?" I will check, like, you know, with the webcam or whatever. Like, this is a near future. The second thing was like, I would love to spend my entire days doing that and buy Mac Minis just for experiences, like a mini lab, for example. Give one a Gmail address and say, like, "Whatever you do, create a business that makes money and continue doing it until you make money. You can create a Stripe account. You can use my LLC if you want." You know, like, and and give just like one task. "Check everything on X, check everything on YouTube, check all the content on the make money. Create a business to make money." And usually, you know, there was always this like meme of, "Make him make a business a billion-dollar business." Make no mistake. I feel like now it's possible. It's no longer a meme.

Yeah, it's not like I'm sure he can do something good, like definitely. So, yeah, that's your IQ or big south. What do you think about it? Yeah, yeah, yeah, I agree. I agree 100%. So, yeah, like it just changed, changed many things. Like, I don't even know how to, like, describe it. Like, like someone else described it in the same way to me. I did not believe them at the start. So until I actually tried it, that's when I knew, like, what, like, it's an AI that can update itself. Like, if it sees that it can't do something, it will update itself and then may, like, it will start doing that. If it, if it doesn't know something, it will actually figure it out and then do it. It won't say something like, "Oh, I don't, I can't do this." Uh, yeah, like, like there is not even a single task where you can say that, "Okay, this, this is something, like, it can't do it." It will figure it out. If there are no actual physical limitations.

Yeah, exactly. Like this morning, when I was working my dog, and I was talking to it, "Create a Notion," and he was like, "I can't." I say, "Of course, you have a Gmail account, you can." And then he actually went on Notion, created an account, took the the password or the the code they send in the email, like creating the Notion. And I, yeah, I think that's, uh, that's totally insane.

So even when I, like, I sent a voice message. I forgot that it can't read, like, voices. So then it went, I went to the web, downloaded a library, like a package which can actually transcribe my voice into text, and then it responded back in voice. So when he did that, like, it's like, it's, it's not talking to me. I can't. So, so yeah, there is, like, there is nothing that it can't do. Like, if you, like, if you give it enough, like, enough, uh, ability to, like, "Okay, I just said that, 'Okay, be proactive. Do whatever you want to do within the, like, constraints I gave it. But be proactive. Don't keep asking me, 'Okay, should I do it? Should I do it or not?'" Then, like, yeah, from then onwards, like, like just this incident, right? It actually downloaded something. It transcribed my audio. It created an audio and then sent it as an audio clip back to me. Yeah, it's crazy.

You know what you're gonna do, Bhanu? We will have the same podcast in one month. Let's do that. I want to know, you know, where you are now, what it built. Oh, Jarvis devolved. Like, I want to have you in one month. That was a special episode. You know, like, usually we go through all the story of the of the guest. I think your story is amazing, and I invite anyone to follow you on X. I will have the link below. You build amazing SaaS, amazing startup. To continue building it, I will put the link also to Mission Control HQ, that AI in the description if you want to give it a try. And yeah, Bhanu, let's do that.

Let's, sounds good, sounds good. At the monthly meeting and talk about it because I'm so excited about this technology, and you are as well, and we are both building on it. So, yeah.

Any last word you want to do to say? Oh, no, no. I think I don't think I'm, like, this has been a great conversation. I think you are just starting out. You haven't even seen, like, all the capabilities that it has. But, yeah, in in one month, you will be able to see how, how it is.

Amazing. Well, Bhanu, have a good day and see you in one year. Yeah, have a good day.