Architecture Pattern
RAG Architecture
A pipeline that grounds an LLM's answers in your own documents via retrieval, instead of relying only on what the model memorized during training.
What it is
Incoming documents are split into chunks, converted into embeddings, and stored in a vector database. At query time, the system retrieves the most relevant chunks and hands them to the model alongside the question, so the model answers from real, current source material rather than guessing from training data.
When to use it
Whenever the answer needs to come from your own, current, or private information — internal documentation, support content, product specs — rather than general knowledge the model already has.
Real tradeoffs
- —Answer quality is capped by how well documents are chunked and indexed, not just by the model's capability
- —Adds real infrastructure (a vector store, an embedding pipeline) that has to be maintained
- —Doesn't change the model's underlying behavior or style — see Fine-Tuning for that
Related
FAQs
Is RAG the same as fine-tuning?
No — RAG grounds answers in retrieved documents at query time; fine-tuning changes the model's underlying behavior through further training. They solve different problems and are often used together.
Have a project in mind?
Tell us what you're trying to automate or build — we'll reply with next steps, not a sales pitch.