LangChain
A framework for building agents and LLM-powered applications by chaining interoperable components and third-party integrations. Maintained by langchain-ai, MIT licensed. It now describes itself as “the agent engineering platform”.
🔗 https://www.langchain.com/ · https://github.com/langchain-ai/langchain
Core Abstractions
Chat models · embeddings · vector stores · retrievers · tools and toolkits · model providers.
The wider ecosystem includes LangGraph (agent orchestration), Deep Agents, and LangSmith (observability and deployment).
Why It Matters Here
Two things make LangChain relevant to this vault beyond being a popular wrapper. Its text splitters (CharacterTextSplitter, RecursiveCharacterTextSplitter) are the de facto reference implementations that most Text Chunking writing benchmarks against. And its memory classes are the clearest taxonomy of Conversational Memory strategies available in code.
Related Tools
- Haystack (deepset) · LlamaIndex — comparable orchestration frameworks
- RAGAS — evaluation, integrates with LangChain
- DSPy — the declarative counter-position to prompt chaining
Related Concepts
Articles
- Conversational Memory for LLMs with Langchain — the memory-class taxonomy
- Chunking Strategies for LLM Applications — uses LangChain splitters throughout