LangChain

LangChain is a framework for developing applications powered by language models. It provides modular components for working with LLMs, including chains, agents, memory, and retrieval systems. Available in Python and JavaScript.

LangGraph extends LangChain with graph-based workflow orchestration for building stateful, multi-actor applications. Features include cycles and branching for complex agent behaviors, state management with persistence, human-in-the-loop approvals, streaming support for real-time outputs, and time-travel debugging. Reached v1.0 in November 2024. Perfect for multi-agent systems, complex workflows, and durable execution patterns.

LangServe helps developers deploy LangChain runnables and chains as REST APIs. Integrated with FastAPI and pydantic for data validation. Automatically infers input/output schemas from LangChain objects, enforced on every API call with rich error messages. Features API docs with JSONSchema and Swagger, efficient /invoke, /batch, /stream endpoints, and playground page at /playground/ with streaming output and intermediate steps. Essentially wraps LangChain chains/agents and turns them into FastAPI endpoints. Hosted LangServe available as part of LangSmith SaaS, directly interfacing with GitHub-hosted code and deploying to containers.

LangSmith is LangChain's platform for debugging, testing, evaluating, and monitoring LLM applications. Trace every step of your chains, run evaluations, and monitor production performance.