Artificial intelligence-native observability startup Braintrust has secured $80 million in a Series B funding round, positioning itself as a critical infrastructure layer for AI models and agents in production. The round was led by Iconiq, with participation from notable investors including Andreessen Horowitz and Greylock.
Braintrust offers an AI-focused observability and evaluation platform designed to monitor the quality of AI models and their outputs. Unlike traditional systems, it tracks model hallucinations, drift, and regressions that lead to inaccurate responses over time. Co-founder and CEO Ankur Goyal highlighted the limitations of existing tools in handling the rapid evolution and complexity of AI systems.
The platform integrates exhaustive tracing tools that capture every step of an AI model's reasoning process, including prompts, tool calls, and cost metadata. It also features automated model evaluation and a playground for testing prompt changes against production data. An AI assistant analyzes millions of traces to suggest prompt improvements and identify hallucination patterns.
Braintrust's proprietary database, Brainstore, is designed for efficient querying of complex AI traces. The company serves major AI-native clients such as Notion, Replit, Cloudflare, Ramp, and Dropbox.
Iconiq General Partner Matt Jacobsen cited Braintrust's deep customer obsession as a key factor in their investment. The new funding will support engineering and go-to-market team expansion, the launch of new observability tools, and regional growth.