Inside OpenAI's Full Agentic Software Factory: From /goal to Self-Healing Production
Summary
On-site deep dive with 7 OpenAI engineering leaders: non-engineering orgs went from ~0% to 90% Codex usage in 4 months; the full pipeline runs human-defines-outcome -> Codex gathers context (docs/Slack/Notion/Databricks/Datadog) -> implements -> agent babysits CI to green -> agentic code review -> agentic deploy with agent-built monitoring -> 'Perf Factory' auto-detects/fixes regressions -> 'Sevbot' triages incidents. A 2-person team completed migrations that needed 6 engineers pre-Codex; PR volume growing ~10x every 6 months.
Why this signal
The most complete, enterprise-scale worked example of focus area #3 (full ideation-to-launch AI-accelerated workflow) available this cycle — a direct benchmark for how far Thomas Engineering's own pipeline could go.
Original email
The Pragmatic Engineer · 9/15/2026
Inside OpenAI's agentic software factory