# OpenAI puts Codex agents on ChatGPT performance regressions, with engineers still approving fixes

> Source: <https://runtimewire.com/article/openai-codex-performance-factory-chatgpt-regressions>
> Published: 2026-11-19 18:22:38+00:00

**Recommendation:** Wire an agent to open a task on each performance alert, investigate and record findings, then route the proposed fix to an engineer for approval before deployment.
More in the playbook "Running AI in production": <https://wpnews.pro/playbooks/production-ops>

# OpenAI puts Codex agents on ChatGPT performance regressions, with engineers still approving fixes

**In just over three months, OpenAI says the internal Regression Workflow ran 1,700 investigations and found root causes in more than 600 cases.**

        By [Ryan Merket](https://runtimewire.com/author/ryan-merket)
        · Published 

OpenAI engineers Okay Zed and Cyrus Forbes disclosed an internal system called the Performance Factory on November 19th, 2026, built to let Codex agents investigate ChatGPT performance regressions, propose fixes, and verify recovery after deployment. OpenAI described the system in an engineering post and a thread on X. The first workflow, called the Regression Workflow, opens a task when a performance alert fires, assigns an agent to investigate, records the findings, and routes a proposed fix to an engineer for review.

Video from the original post on X.…
