Short Answer
For a simple RAG over one document type with one embedding model, raw API calls are simpler. Use LangChain when you have multiple data sources, models, or want a real agent.
Detailed Answer
A basic RAG is: embed the question, query a vector store for the top 5 chunks, stuff them into the prompt, call the model, return the answer. That is 100-200 lines of plain TypeScript or Python with `fetch` and a vector store client. No framework needed. LangChain earns its keep when: you have multiple document types (PDF + Notion + Slack), you want to swap embedding models frequently, you need conversation memory, you want a real agent with tool use, or you want observability via LangSmith. For a single source and a single model, the framework is overhead.
No answers yet
Be the first to answer this question!
Sign in to post an answer