How Braintrust turns customer requests into code with Codex
OpenAI Blog · 2026-05-29
Braintrust CEO Ankur Goyal describes how adopting OpenAI's Codex with GPT-5.5 allowed half the engineering team to migrate to the tool within one month, enabling real-time customer feature request iteration by generating working preview branches in minutes.
Appears in
Extraction
Topics: ai-coding-toolssoftware-development-workflowdeveloper-productivityopenai-codex
Claims
- Half of Braintrust's engineering team adopted Codex within one month of its introduction.
- Codex's speed enables engineers to convert customer feature requests into working preview branches in minutes rather than queuing them in a backlog.
- The speed difference shifts the engineering interaction model from step-by-step guided prompting to writing test cases and letting Codex run autonomously in a sandbox.
- Braintrust uses Codex to ideate on feature requests with customers in real time, shortening the feedback loop between customer need and working implementation.
Key quotes
Codex unlocked our ability to try out customer feature requests in real time. Previously, if someone gave us a feature request, it would enter a backlog and get prioritized later.
With Codex, I've shifted to writing a test that demonstrates a problem, creating a sandbox environment, and then letting Codex run in that environment. This is a novel use case for me, and I can run experiments because of the speed.
The more code we write, the more customer problems we can solve, and Codex is the most effective way to do that right now.