Retrieval-Augmented Generation—the acronym for fetching document text from a database and inserting it into a prompt so a model generates factually grounded answers.

The dominant three-letter acronym of 2020s AI engineering. Dropped continuously in pitch decks and tech keynotes to signal that an app doesn't just guess answers, but reads files first.

• Acronym Flex: Dropping "RAG" continuously in executive calls to signal cutting-edge AI architecture knowledge.
• Vague Scope: Using an acronym that covers everything from a simple 10-line text search script to massive multi-database enterprise architectures.
• Kitchen Reality Check: Like asking a waiter if the soup uses "RAG" (Retrieval-Augmented Gravy).

National Treasure (2000s) scene where Ben Gates retrieves an ancient document from the Library of Congress archives, reads the specific text lines, and uses them to solve the next riddle step.

Fetching relevant documents from a database and giving them to a text model to generate answers.

“Our platform uses RAG to answer employee policy questions using internal company PDFs.”

“We configured our app to search internal policy PDFs first and give those files to the text model to write accurate answers.”