AI 101

RAG

Sketch: a question arrives at a library desk; the right book is pulled from the shelf; then the answer comes with it open.

RAG is a method where an app first looks up relevant documents, then gives them to the model so its answer uses those facts.

Think of it like this: Before the trip, someone pulls the right map from the trunk and holds it up to the windshield. The engine stays the same; it just sees better information.

Source:

https://developers.google.com/machine-learning/glossary#retrieval-augmented-generation-rag (checked Oct 15, 2026)
https://arxiv.org/abs/2005.11401 (checked Oct 15, 2026)

RelatedEmbeddingGroundingContext window Memory Hallucination

Used inBuild a RAG app on Postgres with Supabase and pgvector Context engineering Embedding Fine-tuning Grounding Hallucination Knowledge cutoff Memory RAG knowledge-base Q&A agent for internal docs