Personal retrieval

Personal retrieval is asking your own records a question and getting back an answer that points to the meeting, email, note, or document it came from, so you can open the source and verify each claim before you act on it.

Personal retrieval is asking your own records a question and getting back an answer that names the meeting, email, note, or document each claim came from, so you can open the source and check it before acting. The assistant searches the records you saved, drafts a response, and attaches a link or citation to every statement. Say you ask, "What did we decide about the refund policy?" A plain chatbot might invent a confident-sounding rule. With personal retrieval, the assistant answers from your decision log and shows the dated note where that decision was recorded, so you can click through and confirm the wording yourself.

Builder example

When you build an assistant over your own notes, the difference between a useful tool and a liability is whether each answer can be traced. Tell the assistant to quote the record it used and link back to it for every claim. A support helper that answers "the password reset link expires in one hour" and shows the policy note it pulled from lets a teammate confirm the fact in seconds. The same helper with no source trail forces everyone to take the answer on faith, which is where confabulated details slip through.

Common confusion: Personal retrieval runs over the records you own, while Retrieval-Augmented Generation (RAG) is the broader technique of searching any document set at question time and feeding the matches to the model. Personal retrieval is RAG aimed at your own meetings, emails, and notes, with verifiable sources as the point rather than an optional extra.