You remember making the decision. You remember a line from the meeting. You cannot remember which file contains it. Before recreating the work, start with the smallest detail you can recall.
Search for a distinctive detail
A client name, an unusual phrase or a figure often works better than a broad project title. Search your library for the term, then narrow the results by what you know about the document.
If the phrase is too precise, remove a word. If it returns too much, add a second detail. Try the language that appeared in the original file rather than the language you would use to describe it today.
- Remember a name? Pair it with the project or document type.
- Remember a phrase? Search a few distinctive words from it.
- Remember the project? Return to its space and inspect the linked work.
- Remember the conclusion? Find the source before reusing the answer.
Open the result, not just the preview
A matching title or excerpt is a starting point. Open the document and read the surrounding section. Confirm that it is the right version and that the passage means what you remember.
This matters most for decisions, figures and commitments. A draft budget may look much like the approved one. A meeting note may record an option that was later rejected. Search finds matches; you still need to establish which source is authoritative.
Use AI for a question that spans sources
When you need to compare several documents or describe something you cannot name, an AI request can help. Tell the agent which sources to use and ask it to show where its answer comes from.
Using the launch brief and these meeting notes, find the latest decision about the release date. Show the source for the decision, distinguish it from earlier proposals, and say if the documents disagree.Treat the answer as a guide back to the evidence. Open the cited material and check the context before acting on it. If a document is missing, encrypted or not indexed, do not assume an empty result proves that the information does not exist.
Know which kind of search you are using
Local library search and a cloud AI request have different data flows. Searching indexed text can run on your computer. AI retrieval, connected-app search and other network-backed features may send the query and relevant content to the services involved.
Choose the method that fits the task and the material. You can read more about those boundaries in What stays local. What goes online..
When you find the right source, link it from the project note and add a sentence explaining why it matters. That small piece of context saves the next person from repeating the search.
