Gergely Orosz · 14 min read · Product Teams
Inside OpenAI’s agentic software factory
Read the original on newsletter.pragmaticengineer.com →
OpenAI has largely transplanted Codex and ChatGPT Work into core operations, with non-engineering teams rapidly migrating their workflows to Codex. The company is building a 'software factory' with automated agentic feedback loops, reshaping tooling, code reviews, and how engineering work is done as it scales toward billions of users.
Key Takeaways
- Codex and ChatGPT Work now underpin the majority of workflows across departments, not just engineering.
- The traditional IDE and PR-based processes are fading as Codex usage surges.
- OpenAI operates a 'software factory' using automated, agentic feedback loops to detect and fix issues automatically.
- Internal tooling is being replaced by Codex, with debugging increasingly leveraging Codex capabilities and a focus on efficiency.
- Engineering work is shifting toward broader agency and judgment, with fewer specialization silos as the company scales.
Excerpt
It’s rare to work with an unlimited token budget, but at OpenAI, that’s what all engineers, researchers, finance colleagues, and marketing folks do. Recently, I visited one of the world’s leading frontier labs to find out how OpenAI operates today – and for a glimpse at where software engineering might be headed as a profession. Plenty has changed since I visited OpenAI’s headquarters last year. Within a year, Codex has gone from a “nice-to-have” tool to being the backbone of pretty much everything at the company. To learn more, I talked with seven engineering leaders and engineers there: Venkat Venkataramani…