I know a bootstrapper who’s running a business with zero human employees and 14 AI OpenClaw bots.

Other founders are copying his approach, so I want to show you how it works.

Bhanu Teja P is the founder of SiteGPT, which allows companies to add AI-search to their sites. What he built is incredible.

This is how he manages his bot squad:

What his AI bots have done:

• A product analyst AI bot went through his site's onboarding as a real user would, and found ways to increase conversion

• An SEO specialist AI bot found relevant keywords that should be added to his homepage

• A retention AI bot noticed that users who send 5 messages are 3x more likely to stay subscribed and started improving retention

• Dozens of other tasks — all on their own

Some basics for OpenClaw newbies:

What I’m about to show you is built on OpenClaw, the open source project that founders on X are going crazy for.

You can get it for free, here.

There are loads of videos on how to install it, but I like this one.

After installing it, connect it to your favorite AI model and you’re in business.

Many people are saving money on it by connecting it to their Claude accounts as you can see here.

Why he needs multiple bots:

Most people create one OpenClaw bot and have it do anything and everything.

Bhanu wanted specialists.

So he asked his main agent to create them for him. I’ve been using OpenClaw for weeks and had no idea I could ask for that.

Some of the AI bot roles at the company:

But how does he manage it all?

After Jarvis (his main agent) built the first 4 speciality agents, it was hard to see what they were all doing.

So Bhanu said to Jarvis,

Jarvis created Mission Control, a SaaS that's built for one person: Bhanu.

It looks like this:

How work gets done

Bhanu has an idea
↓
He Telegrams it to Jarvis, his Squad Lead
↓ 
Jarvis adds the task to Mission Control & assigns it
↓
A specialist bot works on it
↓
All other bots see the new task
↓
Other bots add insights to the card
↓
The responsible agent finishes the work

How the bots collaborate

Every project becomes a card on Mission Control’s Kanban board. Jarvis (the main bot) adds it and assigns it.

A sample project:

One specialist bot owns the card and drives it forward, but the system isn’t siloed.

Every 15 minutes, all bots scan the board for new activity. If a bot has relevant expertise—SEO, design, onboarding, analytics—it can attach insights or propose changes. The result is a lead owner with a rotating council of specialists continually improving the work.

Check out these 2 bots, Shuri & Vision, collaborating:

Why do AI bots need a chatroom?!

The chatroom he built for his bots is the kind of watercooler conversations that human employees have.

An agent might chat, "I found this insight that people who send at least five messages are 3x more likely to convert."

Then the retention specialist might pick up the task and say, "I know what to do. I’ll create an onboarding sequence where the entire goal is to activate the user."

And just like that, a new task will be created without Bhanu's intervention.

Chat looks like this:

Cool, but there are problems

The biggest challenge is all the output they’re capable of. They're moving too fast for him, for any human.

He hasn't figured out what he should check before they implement and what he should just let the agents launch on their own.

He’s capitalizing on the growth

So many people are trying to copy this that he realized it could be a new business for him, so Bhanu created a site that launches these bot workforces. It’s here.

Check out the screencast

I have 2 ways for you to see it:

The YouTube Short & the full YouTube video.

Andrew Warner
Chief Curiosity Officer of the Internet

PS Super weird for me to tell you, but I’m too excited to hold it in. This morning I tried climbing the rope at the gym and I did it! It always felt like a thing “other people do.” But the trainer showed me how to do it and encouraged me to try.