LangSmith provides tools for debugging, testing, evaluating, and monitoring LLM applications, integrating seamlessly with LangChain for comprehensive AI agent observability
About LangSmith
LangSmith is the observability and evaluation platform from the LangChain team. It records what an agent or model application does, step by step, so developers can see where an answer went wrong, compare versions and test changes before they ship.
It was built to be neutral about the model and the framework, so it can trace applications built with LangChain or LangGraph and also those built with no framework at all. Tracing is based on OpenTelemetry, and anything that emits that standard can be ingested. Teams use it to build evaluation datasets, run tests against them, and monitor production traffic, and a prompt playground lets them try changes quickly.
LangSmith is a hosted service with options for companies that need to run it in their own environment. It is a common choice for teams already using LangChain, and it is useful anywhere an agent's behaviour needs to be explained and measured.
Key features
- Step-by-step traces of agents and model calls
- Works with LangChain, LangGraph or no framework
- OpenTelemetry-based tracing
- Evaluation datasets and test runs
- Monitoring of production traffic
- Prompt playground
- Hosted, with options for private deployment
Links
More from LangChain
Build resilient agents
The batteries-included agent harness
The agent engineering platform
An Open-Source Asynchronous Coding Agent
Framework to build resilient language agents as graphs.
LangServe π¦οΈπ LangServe facilitates the deployment of LangChain runnables and chains as a REST API, providing features like automatic schema inference, efficient endpoints, and a playground page, with plans for a hosted version for one-click deployments
All content and agents from LangChain
Frequently asked questions about LangSmith
- What is LangSmith?
- LangSmith provides tools for debugging, testing, evaluating, and monitoring LLM applications, integrating seamlessly with LangChain for comprehensive AI agent observability. It is listed on botshub under agent observability and evaluation.
- Who makes LangSmith?
- LangSmith is made by LangChain, a company based in United States.
- Is LangSmith free or open source?
- LangSmith is tagged as open source on botshub. Check the official site for current terms.
- What can LangSmith do?
- Key features include step-by-step traces of agents and model calls, works with LangChain, LangGraph or no framework, openTelemetry-based tracing, evaluation datasets and test runs and monitoring of production traffic.
- What are alternatives to LangSmith?
- Other agent observability and evaluation on botshub include Opik, Arize Phoenix, OrcaReplay and Helicone. See all alternatives to LangSmith.
