LangChainUnclaimed

LangSmith is an agent engineering platform from LangChain. It covers the full agent development lifecycle: building agents, testing their behavior, deploying them, monitoring them in production, iterating on failures, and governing model calls and spend.

The platform is made up of several services. Observability lets teams see step-by-step traces of what their agents are doing. Evaluation scores agent performance and turns expert review into standardized evals. Agent Server is used to ship and scale long-running agents with persistent state. Sandboxes run agent-generated code in isolated environments. Engine monitors production traces, clusters agent behavior into issues, and recommends fixes. LLM Gateway controls model calls, spend, rate limits and sensitive data, and can add model fallbacks.

LangChain's open source frameworks sit alongside the platform. deepagents is for long-running agents on complex tasks, langgraph gives low-level control over agent workflows, and langchain offers quick-start agents with any model provider.

The page says LangSmith is aimed at teams building agents in production, including engineering teams that need to track quality, cost, latency and errors across agent runs. Fleet lets non-technical team members create agents in everyday language.

Pricing has three tiers. The Developer plan is free for one seat, with up to 5,000 base traces per month and then pay-as-you-go. The Plus plan costs $39 per seat per month, with up to 10,000 base traces per month, access to Agent Server and Engine, and unlimited seats. Enterprise pricing is custom and includes self-hosted and hybrid deployment, custom SSO, and support SLAs. Usage beyond plan limits is billed in LangChain Standard Units at $1.00 per LSU.

The page also offers discounted rates and up to $10,000 in credits for VC-backed startups.