Solo Zero: Building a Company on AI Agents

Solo Zero
  • 0.0 (0)
  • Technology
  • Updated Weekly

One person running a whole company on AI agents, and posting the receipts. I'm an ex-Meta staff engineer who walked away from a $900K job to test one thing: can a company eventually run at zero of my hours and still grow, make money, and scale? Four rules. No employees. No investors. Open source what I can, automate what I can. Every episode is a dispatch from the experiment: what worked, what cost me a day, and the numbers. 10K MRR by December or back to job hunting. Building in public, every week.

Episodes

  1. Aug 21

    10,000 Commits This Year. I Reviewed None of Them

    I run a one-person company. AI agents write 400 to 500 commits a day. This month the whole thing collapsed. Runs started and never finished. This video is the autopsy. I dug through the traces, found four infrastructure bugs, and rebuilt the system so agents ship to production without me reading the diff. A planner that defines done with tests. A server-side review gate. A debugger that treats claims as claims. Every screen you see is from production: the traces, the redrive counts, the PR, the costs. Chapters: 0:00 The collapse 0:30 Why I left $900K at Meta 0:52 The autopsy: digging through runs 1:37 Four plumbing bugs 2:18 The deploy thrash 3:43 feature-flow: plans with tests that define done 4:26 preflight: the gate 4:56 A council of models 6:10 caas-debug: claims are not evidence 6:56 The table-delete story 7:52 Open source + waitlist Waitlist for Syndai: https://www.syndai.ai Feature flow: https://github.com/solozerolabs/feature-flow Preflight: https://github.com/solozerolabs/preflight #buildinpublic #AIagents #codingagents

  • 1 Episode

About

One person running a whole company on AI agents, and posting the receipts. I'm an ex-Meta staff engineer who walked away from a $900K job to test one thing: can a company eventually run at zero of my hours and still grow, make money, and scale? Four rules. No employees. No investors. Open source what I can, automate what I can. Every episode is a dispatch from the experiment: what worked, what cost me a day, and the numbers. 10K MRR by December or back to job hunting. Building in public, every week.

Information