OpenAI's agentic software factory detailed — but zero verified cases of fully autonomous long-horizon coding
alex_verem · x · 2026-10-08
Gergely Orosz shares a detailed look at OpenAI's current agentic software factory, based on explanations from multiple OpenAI folks — with the Perf Factory highlighted as especially interesting.
Quoting this, mardehaym pushes back with a sober note: there is still not a single verified example of agents working entirely on their own over an extended period to build complex software that actually works, while humans do it constantly. We can clearly do something agents still can't.
The pairing of fresh inside detail on how OpenAI runs agents with a clear-eyed take on the autonomy gap makes this a useful snapshot of where agentic coding actually stands.
More from coding & agent
- Agent harnesses are crutches; least-privilege action authorization is what matters — andreisavu · 2026-10-09
- Workers Refuse to Write Markdown Files for Agents, Seeing Knowledge Extraction — mattbeane · 2026-10-08
- Fallout: New York runs in your browser, built with Claude Opus 5.5, zero texture or sound files — chrisfirst · 2026-10-08
- When is AI automation worth it? Only for five-minute tasks that keep repeating — gethackteam · 2026-10-08
- Dad dispatches an agent to build an ESP32 music player to quit Spotify for good — natesiggard · 2026-10-08
- Andy Pavlo: AI agents could drive 10-100x database queries, LLMs hit 85% of tuning quality — mattturck · 2026-10-08