Laminar iconLaminar

Open-source observability for AI agents: automatic tracing, browser session replay and clustering of failure patterns

3.4k+ GitHub Stars
Apache 2.0 License
YC S24 Backed

Overview

Laminar is an open-source observability platform built specifically for AI agents rather than single LLM calls. It captures full agent traces automatically from popular SDKs, records browser sessions for browser agents so you can replay exactly what the agent saw, and groups failures into recurring patterns ("signals") so you can find what breaks most often. Traces are searchable, can be turned into evaluation datasets, and PII can be redacted before storage.

Laminar came out of Y Combinator's Summer 2024 batch and raised a $3M seed round in March 2026 led by Atlantic.vc. The platform is Apache 2.0 licensed and can be self-hosted, or used as a managed cloud with a free tier.

The Verdict

Who Should Use Laminar?

Best For

  • Teams running agents in production
  • Browser agents (Browser Use, Stagehand, Playwright)
  • Debugging why long agent runs fail
  • Teams that want an open-source, self-hostable tool

Not Ideal For

  • Simple chatbots with single LLM calls
  • Enterprises that need a long vendor track record
  • Teams already standardized on Datadog for everything

What's Great

  • Built around agents, with automatic failure-pattern detection
  • Browser session recording synced with traces
  • Drop-in instrumentation for Claude Agent SDK, OpenAI Agents SDK, Vercel AI SDK and LangChain
  • Open source and self-hostable
  • CLI and MCP tools to query traces from coding agents

Watch Out For

  • Young company with a small team
  • Few independent reviews so far
  • Less suited to classic LLM app monitoring than Langfuse or LangSmith
  • No public compliance certifications listed

Pricing

View all features & details

Key Features

  • Automatic agent tracing
  • Browser session recording and replay
  • Signals: clustering of failure patterns
  • Full-text search across traces
  • Evaluation datasets from production traces
  • PII redaction
  • CLI and MCP access, Slack alerts

Integrations

  • Claude Agent SDK, OpenAI Agents SDK
  • Vercel AI SDK, LangChain, Pydantic AI, Mastra
  • Browser Use, Stagehand, Playwright
  • OpenHands

How It Compares

Feature Laminar Langfuse AgentOps LangSmith
Open source Yes (Apache 2.0) Yes (MIT) Yes (MIT SDK) No
Self-hosting Yes Yes Limited Enterprise only
Agent focus Core General LLM apps Core General LLM apps
Browser session replay Yes No No No
Free tier Yes Yes Yes Yes

Is this your tool? Add a badge