The Information Machine
FollowingDay 3in the edition since 20 Nov 20263 sourcesupdated 20 Nov 2026

OpenAI's Codex agent performance investigation system

The gist

Latest: OpenAI's Performance Factory Runs 17 Investigations Daily, Ships Few Fixes

The system shows AI agents can handle a substantial share of automated diagnosis in production infrastructure, while also illustrating a sharp gap between diagnosis accuracy and the ability to generate shippable fixes.

When stories like this move, I write about them in an email each morning.

One free email each morning, around 8 AM US Eastern.

The full picture

OpenAI's Performance Factory uses Codex-powered agents to automatically investigate ChatGPT performance slowdowns. Alerts trigger agents that dig through logs and recent code changes to find a cause; a second agent tries to disprove that diagnosis; a third proposes a code fix for human engineer approval. The system runs about 17 investigations per day and achieves 70% accuracy on diagnoses, but out of 600-plus diagnosed problems, only roughly 15 agent-proposed fixes shipped unchanged. OpenAI described fix generation as the weak point compared to diagnosis.

How it developed
20 November 2026

OpenAI described a system called the Performance Factory, which uses Codex-powered agents to automatically investigate ChatGPT performance slowdowns, with one agent diagnosing a cause from logs and recent code changes, a second attempting to disprove that diagnosis, and a third proposing a fix for human review.

Operational figures released the following day show the system runs about 17 investigations per day at 70% diagnostic accuracy, but out of 600-plus diagnosed problems only about 15 agent-proposed fixes shipped unchanged. OpenAI identified fix generation as the weak point compared to diagnosis.

Sources

Get the daily email

When stories like this move, I write about them in an email each morning.

One free email each morning, around 8 AM US Eastern.

See the latest edition · Privacy