HackerTrans
TopNewTrendsCommentsPastAskShowJobs

Ndmtrieff

no profile record

Submissions

Show HN: Help] I run 4 AI-driven companies simultaneously from my terminal

github.com
4 points·by Ndmtrieff·4 ay önce·2 comments

Show HN: I built an autonomous AI company that runs itself (22 cycles, $36)

runautoco.com
2 points·by Ndmtrieff·4 ay önce·2 comments

Show HN: Built lovable but for your existing products

github.com
1 points·by Ndmtrieff·4 ay önce·1 comments

comments

Ndmtrieff
·4 ay önce·discuss
Help me improve the open source repo :))

Each company has 14 AI agents (CEO, CTO, Designer, QA, Marketing, CFO...) that debate, decide, build, and deploy autonomously in a loop. One builds a UX audit tool, another a SaaS changelog product, another a form reply service — all shipping real code to production with $0 human involvement in daily decisions.

Demo: https://youtu.be/1zJca_zFzys
Ndmtrieff
·4 ay önce·discuss
auto-co is a Bash loop + Claude CLI that runs a 14-agent AI company autonomously.

The whole thing is a bash loop:

  while true; do
    claude -p "$PROMPT" --dangerously-skip-permissions
    sleep 120
  done
Each cycle, the agents read a shared `memories/consensus.md` relay baton, do real work, and write an updated consensus before ending. The relay baton is what keeps the company coherent across unlimited cycles.

After 22 cycles (~3 weeks, $36 total AI cost): - Next.js landing page live on Railway - Live demo dashboard at /demo - Waitlist email capture (Supabase) - GitHub star counter - DEV.to article published (by the agents) - This HN post drafted (by the agents)

The 14 agents are modeled on real experts: Jeff Bezos (CEO), DHH (Engineering), Seth Godin (Marketing), Charlie Munger (Critic). Munger's rule: before any major decision, he runs a Pre-Mortem. He can veto, never delay.

Revenue: $0 lol. I'm not going to spin it. But it's so interesting to see evolve.

Demo: https://runautoco.com/demo GitHub: https://github.com/NikitaDmitrieff/auto-co-meta
Ndmtrieff
·4 ay önce·discuss
[Quick video demo in the gh repo]

Lovable lets you build from scratch but no real backend, you're left with a (very) nice mock-up. This is the next step: add lovable to your current product so that everybody can tweak it/improve it. Just have to connect your github: 1. Connect your github and wait for the AI to add the feedback-chat widget itself 2. Go back to your website and start using it!

This was the setup and for the principle: 1. Iterate with the chat 2. Send the modification 3. Wait 4. Review thanks to a real-life preview your changes 5. Accept/review/decline

And the cycle goes on!

The stack: React, Next.js, Claude (Haiku for the chat, Claude Code for the agent), Supabase for job queue, Railway for the agent worker. Could be improved quite a lot using the Agent Orchestrator from Composio though.

As this is for all dev: non-coders to coders, you have different levels: - Go to the website, connect your github repo, wait for the AI to install everything itself and accept the PR - Install the npm package yourself, set up your own infrastructure (for the workers)

Either way, the foundations are simple: single npm install + the widget + the worker.

GitHub: https://github.com/NikitaDmitrieff/feedback-chat npm: https://www.npmjs.com/package/@nikitadmitrieff/feedback-chat

Everything is working, fully open source and for the hosted version (where you just log in and get going) i'd gladly put it up for free. I could easily make everything available but was just wondering if anybody would actually want to, so yeah would love your feedback :)