I built an AI dev pipeline and ran it live on my side project. The Irfrit Loop: ticket comes in → AI writes the code → two independent AI reviewers tear it apart → auto-fix → re-review → QA screenshots → I click merge. Today's run: 7 tickets, 5 PRs merged in ~2 hours. My only job was clicking merge. The reviewers weren't rubber-stamping — they caught real issues: a race condition in a SQL function, a ghost RPC endpoint with no auth check, a missing database migration that silently showed "saved" while failing, and a date column bug that was attributing revenue to the wrong month. Every failure auto-chained to a fix agent. Every fix auto-chained to a re-review. No idle time, no manual coordination. The bottleneck in solo development isn't writing code. It's the review loop. This eliminates it.

0

150

Cover image for Built a personal mission control

Built a personal mission control for my AI agent team 🔥 4 agents running 24/7 on a Mac Mini. CRON jobs, memory browser, infrastructure monitoring, the whole thing. Designed in V0, built with Codex in under 30 minutes. We are so early.

2

3

275

Contra isn’t just about work. Every once in a while, we get to meet up in person: connect with teammates, share ideas, and just have a great time. Earlier this year, we had Camp Contra 2025, and it was an absolute blast. Grateful for this team and the moments we get to share.

4

553

One thing I love about Discover is that it’s built with fairness at its core. And we don’t just say that—we show it. The Discover Score makes the algorithm fully transparent. I don’t think any other platform does this, and that’s why it’s one of our most memorable features. I’m proud of what it represents, and how it was built.

1

524

Henry Ollarves

AI Agent Systems & Platform Infrastructure