Laminar

Laminar is an open-source observability platform built specifically for AI agents, founded by Robert Kim and Dinmukhamed Mailibay and launched through Y Combinator's Summer 2024 batch. With a single line of code it captures everything an agent does — every LLM call, tool use, and function execution — and turns that into traces that show inputs, model reasoning, and sub-agent activity. A feature called Signals uses AI to detect failure patterns in plain English and can send Slack alerts, while an Agent Debugger lets developers rerun an agent from any step with full prior context preserved, and error clusters can be converted directly into evaluation datasets. For browser-based agents, Laminar records session video and syncs it with the trace timeline so developers can see what the agent was looking at when it made a decision. Traces and metrics are queryable with SQL for custom dashboards, and PII can be redacted at scale. The platform is open source under an Apache 2.0 license and can run as a managed cloud service or be self-hosted via Docker, AWS, or GCP, and it is HIPAA and SOC 2 Type II compliant. It integrates with agent frameworks including the Claude Agent SDK, OpenAI Agents SDK, Vercel AI SDK, LangChain, Pydantic AI, and Playwright. Laminar raised a $3 million seed round in March 2026 and counts Browser Use and OpenHands among its customers.