Why we rebuilt LangSmith around the agent

LangChain
Read full story· 7 min
LangSmith has restructured its platform around a unified agent model to streamline the development lifecycle of AI agents across businesses. This rebuild connects building, running, improving, and governing agents in one place, addressing fragmented workflows and governance challenges.

What happened

LangSmith has overhauled its platform to center around a unified agent model, launching the new LangSmith Agent Platform last week. This update allows businesses to manage the entire lifecycle of AI agents—including building, running, improving, and governing—from a single interface. The change addresses previous issues with fragmented workflows and difficulty in tracking and managing agents across different teams and environments. The new system introduces a unique agent record that consolidates all related data and processes for each agent.

Key details

  • The LangSmith Agent Platform was introduced following a Series C announcement.
  • Agents now have a unique record with an owner, a globally unique ID, and a type.
  • The platform supports building agents via a natural language builder, code harnesses, or external frameworks.
  • Dashboards and evaluation criteria are tailored to each agent type, such as coding or customer support agents.
  • LangSmith allows integration with external runtimes and platforms, offering governance and improvement tools without requiring migration.

Why it matters

Companies using multiple AI agents across departments can now manage and monitor them more efficiently, reducing time spent switching between tools and improving oversight. Platform teams gain better visibility into agent ownership, costs, and performance, which helps address governance and compliance requirements. Developers and businesses benefit from streamlined workflows, consistent evaluation metrics, and easier troubleshooting, potentially lowering operational risks and costs.

More on this story


More in Agents