Encyclopedia Evalica / Observability / Active observability
Active observability
/'ak.tiv uhb.zer.vuh'bih.luh.tee/An approach to AI observability where the system continuously analyzes production traces to proactively cluster and surface what matters, so teams get answers without having to ask ad-hoc questions. (noun)
Why it matters
The volume and non-determinism of production AI makes manual inspection and "look at a dashboard, then guess" workflows break down. Active observability aggregates AI-scale data into structured, queryable signal, then automatically finds recurring issues, groups similar behavior, and surfaces the most important issues. This makes it easy to turn observations into datasets, scorers, and measured improvements.
“We moved from reactive monitoring to active observability by surfacing the top failure patterns every day.”
Related Observability terms
- AI observability •
- Alert / threshold •
- Dashboard •
- Data flywheel •
- Deep search •
- Drift •
- Error rate •
- Feedback loop •
- Logs •
- Model drift •
- Online evaluation (production scoring) •
- P50 / P95 / P99 (Percentiles) •
- Sampling rate •
- Service Level Indicator (SLI) •
- Service Level Objective (SLO) •
- Time-to-first-token (TTFT) •
- Token usage / cost tracking •
- Topics
From the docs
Braintrust is the AI observability and eval platform for production AI. By connecting evals and observability in one workflow, teams at Notion, Stripe, Zapier, Vercel, and Ramp ship quality AI products at scale.
Start building