Braintrust

Braintrust has raised an $80M Series B. We're building the infrastructure that helps teams measure, evaluate, and improve their AI products. Don't take our word for it. Hear how @NotionHQ, @Vercel, @Navan, and @billcom use Braintrust to ship quality AI.

Read more → braintrustdata.link/gateway-models

Braintrust Gateway now shows detailed model information for every supported provider. Compare capabilities, context windows, and API types across OpenAI, Anthropic, Google, AWS, and open-source models in one place.

One dataset ≠ one eval. It's possible to run multiple evals on the same dataset to test different measures of AI quality, like factuality, personality, and effectiveness in resolving customer complaints.

@daRubberDuckiee Join us → braintrustdata.link/evals-101-work…

The best AI product managers know whether or not their product is improving. DevRel engineer @darubberduckiee is hosting an Evals 101 workshop built for PMs. Learn how to measure the performance and quality of AI products.

Read more → braintrustdata.link/starter-plan

Starter is a new Braintrust pricing plan with no platform fee. It includes: - 1GB processed data - 10K scores - 14 day retention You can pay for more usage as your AI workloads grow.

Read more → braintrustdata.link/updates-3-13-26

What's new: - Playground bulk optimization: annotate multiple outputs across prompts and let Loop optimize prompts for you - Improved trace and span search: faster, and scoped to the current span or the trace - TypeScript auto-instrumentation: auto-trace all TS LLM calls

Read more → braintrustdata.link/GDPR

For providers subject to the GDPR, the architecture of a platform matters as much as the feature set it offers. Braintrust's platform was intentionally built to separate the control plane from the data plane, making compliance with GDPR feasible.

Trace everything

@daRubberDuckiee Join us → braintrustdata.link/whats-new-march

What's new in Braintrust: Observability and evals work best when built directly into developer workflows. DevRel engineer @daRubberDuckiee demos Gateway, Topics, and trace-level scoring. Learn how teams surface insights across thousands of interactions to improve AI quality.

Read more → braintrustdata.link/dropbox

At @Dropbox, evals are the new PRD. Evaluation scorers are defined before product development begins. Then production traces feed back into new datasets, inform prompt optimizations, and guide fine-tuning decisions, creating a quality feedback loop.

@Replit Watch the full session with @ankrgyl and @pirroh → braintrustdata.link/agent-observab…

Trace fireside chat: Agent observability at scale. @Replit built and scaled AI observability to support rapid iteration without sacrificing reliability.