Braintrust lands $80M Series B to bring observability to production AI
San Francisco-based Braintrust has raised $80 million in a Series B funding round led by ICONIQ, as the company targets one of the most pressing problems in enterprise AI: understanding why models and agents fail after they ship.
Existing backers Andreessen Horowitz, Greylock, Elad Gil, and Basecase Capital also participated. The company said it will use the new capital to expand engineering and go-to-market teams, open additional offices, and build new products.
An “observability layer” for AI systems in production
Braintrust positions its platform as AI observability infrastructure: tooling that helps teams monitor, evaluate, and interpret how AI models and AI agents behave once deployed in real applications. As AI moves from demos into customer-facing products and internal workflows, failures can be difficult to diagnose—especially when multi-step agents generate complex chains of actions and intermediate outputs.
Founder Ankur Goyal said the idea emerged from building internal evaluation tools during his time at Impira and Figma, where diagnosing AI behavior was often slow and highly manual. With today’s agentic systems producing large volumes of session data, the company argues that traditional monitoring tools were not designed to trace issues across long-running processes or explain unexpected results.
Scaling data to explain AI behavior
Braintrust says it built internal database technology to handle the scale and structure of AI-related telemetry, which can reach hundreds of megabytes per session. The company’s broader pitch: observability should be treated as core infrastructure for AI-powered products because models change frequently with updates and new data, raising new reliability and risk concerns.
Customers using the platform include Notion, Replit, Cloudflare, Ramp, and Dropbox. Matt Jacobson of ICONIQ cited strong adoption among leading AI teams and what he described as the company’s customer-driven product development.
Related
