Retrieval-augmented generation
Also used: RAG
Retrieval-augmented generation, or RAG, is a system that looks up relevant source material before an AI model writes its response.
Why it matters to educators
RAG can make an AI response more grounded in current, local, or approved material. That is especially useful when educators need answers based on a district policy, curriculum, or selected collection of resources.
The foundation
A RAG system first retrieves passages that appear relevant to the request, then gives them to a language model as context. The model still generates the final wording, so retrieval reduces—not eliminates—the need to check the answer and sources.
Its quality depends on the source collection, how well the system finds relevant passages, and whether the response clearly cites them. A polished response is not proof that the correct document was retrieved or interpreted accurately.
What this can look like in education
Ask about a handbook
A staff assistant searches the current employee handbook and returns a response with links to the specific policy sections used.
Explore approved curriculum
A teacher asks a curriculum assistant for standards-aligned ideas, then opens the cited district resources before adopting any suggestion.