About

RAG is a way of getting a straight answer out of your own documents. An AI model on its own only knows what it was trained on. It has never read your contracts, your filings, or your policies. RAG closes that gap in two steps.

1. 1 Retrieve. Your question is matched against your documents, and the passages that relate to the prompt are pulled out, searching by meaning rather than by keyword.
2. 2 Generate. Those passages are handed to the model, which answers from them instead of from memory.

The answer is built from your documents rather than from the model's training, which is the whole point: every answer arrives with the passages behind it, so you can check what it was drawn from instead of taking it on trust.

What that lets you do

• Ask a question about a specific document. Point at one contract and ask what its termination clause says, instead of reading forty pages to find out.
• Find the documents most related to an idea. A search for "remote work policy" surfaces the file that says "work from home," even though the words never match.
• Ask across a whole workspace at once. Put a set of documents together and question all of them as one body rather than opening them one at a time.
• Check the answer against its sources. Every answer comes back with the passages it was drawn from, so you can confirm it rather than take it on faith.