Braintrust - The AI observability platform for building quality AI products

# Braintrust > Ship quality AI at scale. Turn production traces into evals, compare prompts > and models, and improve quality with every release. ## For AI agents - [Complete technical documentation with full content (SDKs, APIs, integrations, guides)](https://www.braintrust.dev/docs/llms-full.txt) - [Documentation index](https://www.braintrust.dev/docs/llms.txt) ## What is Braintrust? Braintrust is an end-to-end platform for building reliable AI applications, built around a continuous improvement loop: 1. **Instrument** — Wrap AI providers (OpenAI, Anthropic, Gemini, etc.) or use OpenTelemetry to capture traces from your application 2. **Observe** — Monitor production in real time; find errors, latency issues, and unexpected outputs using search, dashboards, and automated topics 3. **Annotate** — Add human feedback, capture user signals, and build datasets from production traces 4. **Evaluate** — Run experiments to compare prompts and models, write scorers (LLM-based, code, or human) to quantify quality 5. **Deploy** — Ship prompts to production, gate releases on eval results, and monitor impact with online scoring ## Key capabilities - **Scalable logging**: High-throughput trace ingestion with real-time monitoring - **Flexible scoring**: LLM judges, code-based scorers, and human review - **Loop Agent**: AI-assisted optimization — auto-generates improved prompts, scorers, and datasets - **Brainstore**: Purpose-built database for querying complex AI traces - **MCP integration**: Connect your IDE and agents directly to your AI stack - **Release gating**: Block bad releases before production with automated evals ## Security & compliance SOC 2 Type II, GDPR, HIPAA. SSO/SAML, granular permissions, hybrid deployment. ## Links - [Docs](https://www.braintrust.dev/docs) - [API reference](https://www.braintrust.dev/docs/api-reference) - [Cookbook](https://www.braintrust.dev/docs/cookbook) - [Blog](https://www.braintrust.dev/blog) - [P