← All posts

Search

Local full-text search for meeting transcripts

Find decisions, names, and exact phrases across meeting transcripts without sending each search query to an AI service.

Meeting archives become less useful as they grow unless the content inside them remains searchable. Titles help when you remember the name of a meeting. They do not help when you only remember a phrase, concern, decision, or action item from the conversation.

Full-text search solves a different problem: find the words inside the notes, not only the label attached to the session.

What meeting search should include

A complete search should look across:

  • session titles;
  • generated summaries; and
  • full transcripts.

Each source serves a different kind of memory. Titles help with known meetings. Summaries help with main themes and outcomes. Transcripts preserve details that may never appear in a condensed summary.

Quendian searches all three across the completed meeting archive.

Why context matters

A matching word by itself is often ambiguous. The surrounding sentence or passage tells you whether the result is the decision you need, a passing mention, or an unrelated use of the same term.

Quendian shows context with search results. Transcript matches can also include speaker and timestamp information when available. Selecting a result opens the relevant meeting and takes you back to the matching passage.

The goal is not simply to prove that a word exists. It is to reconnect you with the source quickly enough to use it.

Local search and AI questions are not the same thing

An AI assistant can be useful for broader questions, but it introduces interpretation. It may summarize several passages, infer a relationship, or produce an answer that needs verification.

Full-text search is more direct. You enter a word or phrase and inspect the matching source yourself.

Quendian uses localized indexed search on the Mac. Search terms, matching transcript passages, and summaries are not sent to Quendian Cloud or a configured AI provider just to return the results.

This makes local search useful for both speed and control. It also provides a source-first foundation for future evidence-backed questions across the archive.

Examples of useful meeting searches

Search works best when you remember some of the language used in the conversation:

  • a customer’s project name;
  • a feature that was postponed;
  • a phrase from a pricing decision;
  • a person’s name;
  • an error message discussed during troubleshooting;
  • a deadline or month;
  • a contract term; or
  • a distinctive action item.

If a broad search returns too much, add another specific word from the same context.

New notes become searchable automatically

Quendian adds completed transcripts and summaries to its local index as they finish. You do not have to maintain a separate search database or manually prepare each note.

The index is designed to keep search responsive across thousands of sessions and millions of words. The user-facing archive remains on the Mac.

Search should lead back to evidence

The most important design choice is that a result remains connected to the source passage. Meeting notes often contain commitments, disagreements, and details that lose meaning when separated from who said them and when.

Local full-text search gives you a fast path from partial memory to inspectable evidence. Instead of guessing which meeting contained a decision, search the archive and return directly to the conversation.

For a feature overview, read Introducing full-text search in Quendian.