LangChain
The original LLM application framework. Provides primitives (models, prompts, parsers, retrievers, tools, chains) that compose into agent and RAG systems.
Last updated: April 26, 2026
Definition
LangChain was the first widely-adopted framework for building applications around LLMs. It provides a large library of integrations (every major LLM provider, vector DB, document loader, and tool) and a set of composition primitives: prompt templates, output parsers, retrievers, tools, and chains. It is the foundation that LangGraph and LangSmith sit on top of. As of 2026, LangChain is most useful as the integration layer (when you need to connect to a specific vendor and want a pre-built adapter) rather than as the orchestration layer (LangGraph took that role).
When To Use
Reach for LangChain when you need a quick integration with a specific vendor or document loader. For orchestration use LangGraph; for observability use LangSmith.
Related Terms
Building with LangChain?
I've shipped this pattern in real production systems. If you want a second pair of eyes on your architecture, that's what I do.