framework
LangChain
LangChain is a broad framework for building LLM applications with chains, tools, retrieval, and integrations used alongside agent orchestrators.
Best for
- Composable LLM app building blocks
- RAG pipelines and tool integrations
- Teams standardizing on a large connector ecosystem
Pros and cons
Pros
- Huge integration surface
- Familiar abstractions for many teams
- Works with LangGraph for agent control
Trade-offs
- Abstraction layers can obscure behavior
- Not all apps need the full framework
Overview
LangChain provides building blocks for LLM applications: prompts, tools, retrievers, and integrations. For modern agent control flow, many teams pair it with LangGraph.
Fit signals
Use LangChain when ecosystem connectors and shared abstractions speed you up more than they cost in complexity.
Production note
Prefer thin, explicit modules for critical paths. Framework magic is fine for prototypes; production needs inspectable tool and retrieval boundaries.