Agents In Use
What’s new todayLast update
82 use cases added today

How to use Grok Bot as a CTO that runs a team of coding bots

Ray Fernando hands Grok Bot a GitHub repo to run as CTO. It spins up cloud agents, hires child bots for PRs, backend and auth, and he warns it once burned two billion tokens in a day.

Source published x.comAgent: Grok Bot
OUR ADAPTATION

Suggested prompt

Suggested prompt written by Agents In Use, not the source author. Replace bracketed inputs and review tool permissions first.

You are the CTO for [repo]. Break the backlog into tasks, start cloud agents for them and follow each PR to merge. If you are overloaded, propose child bots with clear owners and wait for my approval.

Evidence at a glance

Sourced account
Source poster / publisher
Ray Fernando (@RayFernando1337)
Original account
Read the source ↗
Published
Source link checked
Reported result
Ray says the bots take him out of the coordinating seat, and reports one day above two billion tokens before he changed the setup. No PR counts are given.
Human involvement
Ray reads the bots' threads view-only and adjusted the setup after the token spike.

A linked account supports attribution. It does not mean we independently reproduced its result.

Report an issue with this caseOpen correction email

Opens your email app with the case link included. Nothing is sent automatically.

How the agent did it

  1. He gave Grok Bot a GitHub repo and told it to run the show as CTO.
  2. It spins up cloud agents and follows the pull requests.
  3. A second prompt told it to hire child bots that report to it.

@RayFernando1337 wrote that he turned Grok Bot into his CTO. He hands the bot a GitHub repo and tells it to run the show. It spins up cloud agents away from his own machine, follows the pull requests, and uses a rigorous working mode only when a task needs it.

His second prompt tells the bot it is overloaded and should hire child bots that report to it. After that, he says, the bots talk to each other rather than to him. One lands the PRs, one works the Convex backend and one owns authentication, and he reads their threads in view-only mode.

He gives two honest warnings. First, it burns tokens. He says he went past two billion tokens in a single day because he told the bot to use a heavyweight planning plugin for everything, and the fix was to save it for hard tasks. Second, it does not replace engineers. In his words, it takes him out of the coordinating seat, which is a different claim.

The full walkthrough is in an attached video with timestamps, and this entry relies only on the post text. He does not give figures for PRs merged.

Who it's for: Solo developers who want to stop coordinating coding agents by hand.

Prompt
Suggested prompt (written by Agents In Use, not the author's)

Source: X · Ray Fernando (@RayFernando1337)

Frequently asked questions

What do the child bots do?

Ray says one lands the PRs, one works the Convex backend and one owns auth, and they talk to each other instead of to him.

What is the main cost warning?

He went past two billion tokens in one day by using a heavy plugin for every task, and fixed it by saving that for hard tasks.

More details

Remaining steps

  1. Child bots handle PRs, the Convex backend and auth while he reads view-only.

About this account

The CTO bot delegates to cloud agents and child bots that coordinate with each other rather than with Ray.

Platform: X. Attribution: https://x.com/RayFernando1337/status/2090588423350616494. Read the original post text, author and publication date through the X API on 2026-10-19. Found via the grokbot.dev use-cases page; every fact here comes from the original post, not grokbot.dev's summary. Prompt: Suggested (written by Agents In Use, not the author's).

Credit: Ray Fernando (@RayFernando1337) — X

Source published

Added to Agents In Use

Source checked

Tools and topics

Grok Bot · GitHub · Cursor cloud agents · Convex

Coding agents · Engineering management · Token costs

Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.