What is Laminar?
Laminar is an open-source observability platform for AI agents that provides tracing, signal extraction, and evaluations at scale.It captures full trace context and browser session replay, linking spans, tool calls, and screen recordings for diagnosing agent failures.
Developers can rerun executions from any step with preserved context, tune system prompts, and use a browser-based agent debugger to reproduce and fix issues.Built-in signal extraction and AI-assisted trace summarization let teams define custom event schemas and analyze large volumes of traces for recurring failures.
A SQL API and editor enable platform-wide queries to feed Evals datasets and build custom dashboards correlating tokens, latency, and user sessions.The Evals SDK supports automated evaluations and regression monitoring across agents and datasets.
Laminar is Apache 2.0 licensed, integrates with frameworks and SDKs such as LangChain, Anthropic, Playwright, and Browser Use, and can be self-hosted via Docker or Helm; it is implemented in Rust for high-performance handling of large data volumes.
Laminar pricing Freemium
Verify on the official pricing page.
View plansLaminar user reviews
Would you recommend Laminar?
Laminar's key features
-
Span-level agent tracing with full trace context and SDK/framework integrations
-
Interactive agent debugger with step replay, rerun-at-step and live prompt tuning
-
Signal extraction and analysis to detect and aggregate events across millions of traces
-
Evals SDK for automated evaluations using custom datasets and metrics
-
Platform-wide SQL querying and custom dashboards for spans, traces, and eval data
Laminar use cases
-
Debug and optimize complex AI agent workflows using Laminar by replaying full trace context and browser sessions, rerunning failed executions to identify root causes (prompt, logic, or connector issues), and validating fixes end-to-end
-
Extract and summarize large-scale signals and traces with Laminar to surface performance regressions and user-facing errors, then use built-in Evals and SQL API to prioritize improvements and generate actionable reports for engineering and product teams
-
Build observability dashboards and run ad-hoc SQL analytics over agent traces with Laminar to correlate events and session replays, monitor SLAs, create alerts for anomalies, and measure the impact of model or pipeline changes over time
Who is it for?
-
Software developers
-
Machine learning engineers
-
Product managers
-
Devops engineers
-
Data scientists
