Pay for Yourself or You're Fired: A One-Person Company Powered by Three Bots

Elon Musk said Grok Bot can pay for its own subscription. Someone took it seriously and set a rule for the Bots: pay for yourself or you're fired.

Khairallah AL-Awady posted this setup on X. One group chat, three Bots, each with its own memory.

Scout scans all posts in your field on X and Reddit and doesn't touch your drafts. Builder turns every idea it brings back into three things: a reply, a post, and a landing page. Closer wakes up every night and goes through stuck tasks, dead ideas, and drafts no one approved.

Three rules run alongside:

You approve a draft, it ships. The company learns from what sells, not from what it reads.

Drafts don't have a price tag; Closer asks you, doesn't guess.

If an idea hasn't moved in 7 days, kill it.

A one-person company can be busy for months with zero revenue. This setup is aimed squarely at that.

He also posted the prompt:

Build me a one-person company with three bots in one group chat. Keep my raw ideas untouched in ideas/, keep drafts/ updated yourself, ship nothing without my yes and run a nightly check for stuck tasks. Show me the structure first. Do not create anything until I confirm

To start, pick just one niche you actually understand; don't chase every trend on X.

Grok Bot is an app made by xAI. A Bot is an AI teammate that shares a persistent cloud computer, works for you on a schedule, and keeps working even when your laptop is closed. The subscription is included in The X Subscription: Premium starts at $30/month, Super is $100, Ultra is $200, and Plus at $8 does not include it. Usage shares a weekly pool with Grok and Cursor.

Most people treat it as a smarter search box. They ask questions, give one giant Bot all the work, and set a routine to run every 15 minutes. That's 96 times a day, each one drawing from the same weekly pool—even if it finds nothing.

People actually building with it do the opposite: one hard-coded responsibility per Bot, evidence before done, and run only when something changes.

xAI's docs break building into seven pieces: Bot, Computer, Connectors, Skills, Routines, Pages, Approvals, and Vault. A few pitfalls are worth knowing up front. All Bots share the same computer. If one logs into a site, every other Bot can get in. Don't treat a Bot as a security boundary. Put persistent things in /workspace; temp directories and installed packages may disappear when the computer restores. If a connector exists, don't use the browser; the plugin is cheaper and more reliable. Routines have no dry run; a test run is a real run. Every run costs credits, even if it does nothing.

Lingxi Li on xAI's engineering team uses six Bots to manage more than 200 cloud agents. Five engineering Bots each own one domain, and one ops Bot named Jenny is the only one on the team that doesn't write code. Every day at 5 a.m. she has a one-on-one with each Bot, goes through the playbook, and digs out blockers. Every PR must include evidence; screenshots must show before and after. At 3 a.m. there's an audit routine that clears dead code and checks load time and package size.

Ryan Perry uses six Bots to run the mobile game Rank'em. The analyst Bot is the only one allowed to announce findings. The creative Bot turns findings into ad creative but never buys media. The data showed players were couples playing together, not large groups. They switched to a couples ad and CPI dropped from $15 to $1. The numbers are his own account, but the structure is the point: every spending decision waits for his approval.

Strip both setups down and three habits are the same: one hard-coded responsibility per Bot; evidence before done; run only when something changes, not by polling.

Back to the trio. Scout finds, Builder builds, Closer checks. Closer's nightly check is a scaled-down version of Lingxi's 3 a.m. audit. The difference is that it even counts drafts you've delayed approving as "stuck tasks."

One judgment needs to be clear. Bots don't decide what's worth building. They can't tell a useful tool from a pretty toy. That judgment is still human work, and no one has ever been able to do it for you.

So most people type "build me something" and then wonder where the hype is. A few write a sentence describing the outcome, run it manually once, save it as a skill, set boundaries, automate it, and then read the run logs. Same Bot, same $30—the difference is here.

Tonight, pick the first thing you want to build. Write that sentence first.

Original link: https://x.com/eng_khairallah1/status/2115873637366637016