# How Pylon uses Braintrust

**~10 → 0 mins** Engineering time to debug an issue

As published on [braintrust.dev](https://www.braintrust.dev/customers/pylon). Captured by usedby on 2026-10-06.

- Company: [Pylon](https://www.usedby.ai/companies/pylon.md)
- Tool: [Braintrust](https://www.usedby.ai/tools/braintrust.md)
- Teams: AI Team, Engineering, AI on-call team

## What the story says

Pylon, an agentic B2B support platform, uses Braintrust to test every AI prompt through playgrounds with curated datasets, enforced as a CI requirement. It also uses Braintrust for production observability, building eval datasets from production edge cases, LLM-as-a-judge scoring, and automated root cause analysis of on-call bugs via the Braintrust MCP and CLI.

Summary written by usedby from the source page, in English. The figures are those of Braintrust and Pylon, not ours.

> Previously an on-call engineer had to spend ten minutes actively digging into an issue. Now they run this command, set it and forget it, go grab a coffee, and come back to the root cause totally figured out: a timeline, turn by turn, of what was happening and what went wrong.

- **100%**: “100% enforced, playgrounds by default”

> This discipline is what Braintrust was built for, which is not just logging and observability, but improving quality by bringing it into our everyday development process.
>
> Fred Zhao, Software Engineer, AI Team

## What usedby checked

We compared the story with its live page on 2026-10-06.

- Checked: the figure ~10 → 0 mins is on the page; its label is our wording.
- Checked: the passage quoted above is copied word for word from the page, near the name of Pylon.
- Checked: each number in our summary is printed on the page.
- Checked: Pylon uses Braintrust. Confirmed line. Latest check across sources: 2026-10-06.
- Not checked: the result itself. We quote it; we did not measure it.

---
Source: https://www.usedby.ai/case-studies/pylon-braintrust · How we check: https://www.usedby.ai/methodology