Langtrace

Langtrace is an open-source observability and evaluation platform for AI agents, built to help teams turn AI prototypes into scalable, production-ready systems.

Observability for AI systems

Langtrace collects and visualizes the operational data behind your agents so you can understand behavior from request to response.

• Centralized logs, traces, and performance metrics for agent runs
• End-to-end visibility into model and prompt behavior
• OpenTelemetry (OTEL) support to fit into existing monitoring stacks

Quality and safety evaluation

Alongside observability, Langtrace adds evaluation workflows to measure how well your agents perform and where they fail.

• Track response quality, stability, and latency over time
• Compare prompt and model versions to spot regressions
• Identify bottlenecks and potential safety risks systematically

Built for production teams

Langtrace is designed for teams shipping AI products to production, including developers, MLOps, and product managers who need reliable, auditable AI services.