The meeting you're looking for is a text file, not a database query

You remember that someone agreed to push the launch back a week. You don't remember which of the four calls you had that week it happened in. This is the moment a meeting archive either saves you five minutes or costs you twenty, and the difference comes down to one thing: whether what was said is stored as searchable text somewhere you can actually search it.

What's actually on disk

Coii AudioNotes runs transcription locally on every recording, and nothing about that transcript is uploaded anywhere — it exists as a file, on your machine, the same as a document you typed yourself. The minutes generated from it are the same: text, local, yours. That's the part worth understanding before the "how," because it's what makes searching past meetings possible at all without a subscription-backed search index living on someone else's server. A plain text file on your Mac is something Spotlight already knows how to find, and this is one of the practical benefits of an offline-first meeting recorder that a cloud tool's own marketing rarely mentions: search that doesn't depend on a service being reachable.

Three ways to find a meeting again

Browse the list by date. The app keeps your recordings as a list, which is the right tool when you remember roughly when something happened — "it was the Tuesday before the offsite" — more than what exact words were used. Scanning a list of a dozen entries by date is often faster than typing a search term you're not sure is right.

Search by a word you remember being said. Because the transcript is stored as ordinary text, macOS's own Spotlight search — the same one you use to find a document or an email — can find a word or phrase inside it. If you remember the word "renewal" came up, or a project's codename, or a person's name that was said out loud, that's your search term, and it works the same way searching for that word in a Pages document would.

Search by what the meeting was about, not what was said word for word. The generated minutes condense a meeting down to a handful of decisions and action items, in language closer to how you'd actually describe the meeting afterward than the raw back-and-forth of the transcript. If your memory is closer to "the call where we decided not to raise prices" than to the exact sentence someone used, the minutes are often the faster text to search than the full transcript.

Why this works differently from a cloud notetaker's search

A cloud-based notetaker with a search bar is running that search against a database on a server, which means the search only works when the service is up, your account is in good standing, and you're logged in from wherever you happen to be. It's also a search across everything that was ever uploaded — including, per several of those tools' own feature lists, data that may be used to improve their models unless you've found the setting to opt out.

Searching text on your own Mac is a smaller feature by comparison, and it comes with a real limitation: there's no search across multiple Macs, because there's no shared account or sync to search across in the first place. Each of the three Macs on a licence has its own local archive. What it trades that mobility for is straightforward: nothing you're searching through ever had to leave your machine to become searchable, and the search still works with no internet connection at all, because there was never a server in the loop.

Building an archive that stays easy to search

The habits that make old meetings findable later are mostly about not deferring small bits of upkeep, rather than anything elaborate:

Rename speakers as you go, not in a batch later. A transcript with real names in it is more useful to search — "what did Priya say about the renewal" is a search you can only run if her name replaced the placeholder label at some point. This is the same rename step covered in telling speakers apart in more detail, and it pays off doubly here.

Don't let recordings pile up unlabeled in your head. The app's own list is organized by date rather than by a title you type, so if you rely purely on memory of "it was sometime in March," narrow the search by browsing that window before typing anything.

Keep the transcript, even after you've read the minutes once. The minutes are a compression of the meeting, and a compression necessarily drops detail. If you think there's a chance you'll need an exact quote or a specific number later, the full transcript is what preserves it — and it's what a text search actually searches, in more detail than the summary alone gives it.

Think about how you'll phrase a future search while the meeting is still happening. If you know a number or a name matters, saying it clearly out loud rather than trailing off gives the transcript a cleaner version of that word to search for later. A term that was mumbled and transcribed slightly wrong is harder to find again than one that was stated plainly.

Who ends up relying on this the most

Anyone whose job is a string of similar-sounding meetings feels the lack of search fastest. Running a private practice means months of client sessions that all touch on similar topics, where being able to find "the session where we first discussed the sleep schedule" matters more than being able to browse chronologically. A local-first workflow built around a Mac rather than a shared team workspace has the same shape at smaller scale: a personal archive that only you ever search, growing week over week, with no admin panel and no one else's account involved in finding something in it.

Contrast that with a general meeting notes app built around a shared team inbox — search there is solving a different problem, namely finding something a colleague recorded that you never sat in on. That's a real feature and a real use case, and it's simply not the one a single-user, local-only archive is built to answer. Knowing which problem you actually have determines which of the two you should reach for.

Why local search is a smaller feature that's always on

It's worth naming plainly what this is and isn't. This isn't a purpose-built search engine with filters for date range, participant, or meeting type — it's ordinary file search, applied to ordinary text files that happen to be transcripts and summaries. That's a smaller feature than a dedicated search interface a cloud tool might build for itself, and it's honest to say so rather than implying otherwise.

What it trades in dedicated features it makes up for in always being available. There's no separate search index to keep in sync, no possibility of the search being down while the rest of the app works, and no distinction between "recent" and "archived" meetings that changes how searchable something is — a transcript from three years ago is exactly as searchable as one from this morning, because both are just text files sitting in the same place on the same disk. For a personal archive that one person searches occasionally, that simplicity tends to matter more day to day than a filter panel would.

A worked example

Say it's been three weeks since a call where a vendor agreed to a discount, and finance is now asking for the number in writing. You don't remember which day the call was, only that it was sometime after a product launch and before a board meeting. Start with the list view, narrowed to that window — most weeks only have a handful of external calls in them, so the list alone might get you there in ten seconds of scanning titles and dates.

If it doesn't, open Spotlight and search for the vendor's company name. Because that name was almost certainly said out loud during the call — introductions, or a reference to "the [vendor] contract" — it's sitting in the transcript text, and Spotlight will surface the file the same way it would surface an email mentioning the same company. From there, opening the transcript and searching within it for "discount" or "percent" narrows a forty-minute conversation down to the one exchange that matters, in under a minute total.

This is the pattern worth internalizing: list view for "when," full-text search for "who or what was mentioned," and the minutes for "what was decided." Which one gets you there fastest depends on which detail you happen to remember first, so it's worth trying whichever comes to mind rather than assuming there's one correct way to search.

What doesn't carry over from a cloud tool's search

If you're used to a cloud notetaker's search bar, a couple of differences are worth setting expectations around before you switch. There's no fuzzy semantic search that understands "the call about pricing" means something different from "the call where we discussed cost" unless one of those specific phrases was actually said — Spotlight matches text, not intent, so a search is only as good as the words it's built from. There's also no search that spans a whole team's meetings at once, since there's no team workspace to span; a colleague's recording of the same client call lives on their Mac, not yours, unless you've separately shared the file with each other. Neither is a defect exactly — they're the direct consequence of an archive that has no server behind it to run that kind of query against, which is the same property that keeps every meeting in it private to the machine that recorded it.

A note on how much this archive actually grows

A year of weekly meetings, each transcribed and summarized, adds up to a surprising amount of text — often more searchable words than a person writes in emails over the same period. That's a good thing for search specifically: more text means more distinctive phrases to search on, and a distinctive phrase is easier for Spotlight to zero in on than a common one. It's worth remembering this the first few times a search comes back with more than one plausible result — narrowing by date, or adding a second word from the same conversation, usually gets to the single file you actually meant faster than reading through every match.

Keeping the archive itself safe

Search only helps if the files it's searching still exist, and because everything lives on one Mac rather than a server with its own backups, that responsibility sits with you rather than a vendor. Time Machine or any other backup you already run for the rest of your Mac covers this archive the same way it covers your documents and photos — there's nothing special to configure, because a transcript is just a file like any other on the disk. It's worth checking, once, that whatever backup routine you rely on actually includes the folder these recordings live in, rather than assuming it does. Losing a drive is a much smaller problem when the meeting archive on it was backed up the same way everything else was.

When the meeting genuinely isn't findable

If a recording never happened — the app wasn't running, or the meeting was audio your Mac genuinely couldn't hear, like a phone call taken away from the room — there's nothing to search, because there's no transcript. That's a limit worth naming plainly rather than working around: this only finds what was actually recorded and transcribed in the first place. For everything that was, the answer to "where did we discuss this" is one Spotlight search away, on a Mac that never had to ask anyone else where the file was.