Technology RadarTechnology Radar

LangChain

rag
This item was not updated in last three versions of the Radar. Should it have appeared in one of the more recent editions, there is a good chance it remains pertinent. However, if the item dates back further, its relevance may have diminished and our current evaluation could vary. Regrettably, our capacity to consistently revisit items from past Radar editions is limited.
Assess

LangChain is one of the most widely-adopted frameworks for building LLM-powered applications — particularly RAG pipelines (connecting LLMs to your own documents and data). However, for agent orchestration, its successor LangGraph is now recommended.

Buy vs Build

LangChain is a build framework. There is no "buy" version — it's open-source Python (and JavaScript) code you include in your application.

Why It's in Assess (Not Trial)

LangChain has 80K+ GitHub stars and remains extremely useful, but the team has been clear: use LangGraph for agents, not LangChain. The reasons:

  • LangChain's agent abstraction is verbose and hard to debug
  • LangGraph, built by the same team, provides a cleaner mental model for stateful agent workflows
  • LangChain's "chain" abstraction was designed for sequential pipelines, not the branching, looping nature of modern agents

Where LangChain is still excellent:

  • RAG pipelines — connecting LLMs to PDFs, databases, websites, and knowledge bases
  • Document Q&A — build "chat with your docs" applications quickly
  • Integration breadth — 600+ integrations with data sources, vector stores, and tools
  • LCEL (LangChain Expression Language) — declarative syntax for composing retrieval + generation pipelines

When to Use LangChain vs LangGraph

Use Case Recommended Tool
RAG / document search LangChain
Chat with your docs LangChain
Multi-step agent workflows LangGraph
Complex branching logic LangGraph

Getting Started

pip install langchain langchain-anthropic langchain-community

from langchain_anthropic import ChatAnthropic
from langchain_core.prompts import ChatPromptTemplate

chain = ChatPromptTemplate.from_template("Explain {topic} to a beginner") | ChatAnthropic(model="claude-sonnet-4-6")
result = chain.invoke({"topic": "vector databases"})

Key Characteristics

Property Value
Language Python and JavaScript/TypeScript
License MIT
Integrations 600+ data sources and tools
Provider LangChain Inc.
Website python.langchain.com
GitHub langchain-ai/langchain

Further Reading