Reference · Glossary
LangChain
A popular **Python/JS framework** for chaining LLM calls — prompts, retrievers, tools, agents, and memory in composable pipelines.
When to use
Prototyping RAG apps, agents, and multi-step workflows without wiring every API call by hand.
When not to
Simple one-shot API calls — plain SDK code is often clearer with less abstraction.
Example
chain = retrieval_chain | prompt | llm
answer = chain.invoke({"question": "What's our refund policy?"})