Braintrust
Free tierThe AI observability platform for building quality AI products at scale
Key strengths
Developer Documentation
SDKs & Instrumentation
Braintrust provides native SDKs for Python, TypeScript, Go, Ruby, C#, and more. Adding tracing to an existing AI application requires just a few lines of code. All SDKs support span-level instrumentation, meaning you can capture sub-calls within agents, tool invocations, and retrieval steps individually.
Brainstore
Brainstore is Braintrust's purpose-built database for AI observability data. It is engineered for the large, nested structure of AI traces and outperforms traditional databases on full-text search speed, write latency, and span load time. It supports millions of trace queries with low latency and powers custom views, annotation interfaces, and faceted filtering.
Evaluations & Scoring
Evals can be scored via:
- LLM-as-judge — Use a model to score outputs against a rubric
- Code-based scorers — Write deterministic functions for structured outputs
- Human annotation — Build custom annotation UIs (task-specific trace views) with no frontend work
Datasets are versioned and can be populated from production traces in one click (Trace to Dataset).
Integrations & MCP
Braintrust is framework-agnostic and integrates with any AI provider or SDK. Its MCP server connects your coding agent (e.g., Cursor, Claude) to Braintrust so you can query logs, trigger evals, and update prompts from your IDE. See the integrations page for a full list of supported AI providers and frameworks.
Security & Deployment
- SOC 2 Type II certified, GDPR & HIPAA compliant
- SSO/SAML, RBAC, and granular permissions out of the box
- Hybrid deployment: run the Brainstore data plane on your own infrastructure for data residency requirements
