Agent building & orchestration
LangChain's runtime for running agents in production alongside the open-source LangGraph library. It provides durable execution, threads and runs, and middleware that pauses tool calls for human approval, edit or rejection. Governance beyond approvals and tracing is developer-configured rather than an administrative control surface.
hybrid · generally available · Research snapshot 2026-09-06
Visit the official product source ↗Agent building & orchestration · Observability & traceability
Useful conversation with: AI engineering lead, Platform engineer, Head of AI.
Show me an agent deployed on LangSmith Deployment pausing on a database-write tool call, and a reviewer editing the arguments before it resumes.
Support labels reflect the supplied research. Documentation and vendor claims are not independent product tests. “Not established” means the researcher did not find support; it does not prove a capability is absent.
Documented by provider
Human-in-the-loop middleware checks each tool call against a configurable policy and interrupts execution so a human can approve, edit, reject with feedback, or respond directly, with graph state persisted so the run resumes later.
Limit: Approval policy is authored in code by developers; the docs do not describe an administrator-facing approval console or segregation of duties.
Documented by provider
The Agent Server runtime provides durable execution, streaming and horizontal scaling with assistants for configuration, threads for state and runs for workloads, and can run in LangChain Cloud, a standalone server or self-hosted.
Limit: Page does not document tenant isolation guarantees or audit-log retention.
Documented by provider
Deployment is framework-agnostic, supporting LangGraph, LangChain, Deep Agents, Google ADK, Claude Agent SDK, Strands, CrewAI and AutoGen agents.
Limit: Managed Deep Agents are described as available through a private preview API, so not all listed paths are generally available.
Not established in this research
Administrative governance features such as role-based permission scoping, audit logging and tenant isolation are not established by the pages reviewed.
Limit: Absence of evidence, not evidence of absence; LangSmith may document these elsewhere.
Listing does not imply partnership, supplier status, a working DutyGraph integration, or a compliance certification.
Suggest a correction