Enhance & Auto-Enhance
Turn a raw transcript into a structured note, automatically after each recording or on demand.
Enhance turns a recording and whatever you typed during it into a clean, structured note. It's the step between "I have a transcript" and "I have notes I'd actually reread."
By default it runs on its own the moment you stop recording.
Two Ways It Runs
This distinction explains most of Enhance's behavior, so it's worth two minutes.
| What it reads | What it does to the note | |
|---|---|---|
| On a recording — automatically when you stop, or via the per-recording button | That one recording's transcript, plus your notes | Replaces the note the first time, then appends after that |
| On the whole note — the chevron menu next to the wand | The note and everything attached to it | Always appends a section |
The first row is the normal path and the one this page is mostly about. The second is there for when you want to re-run Enhance over a note by hand.
What You Get
Enhance reads the transcript and decides what kind of recording it was, then shapes the output to match.
Several people talking, back-and-forth, decisions being made:
- Summary — what happened
- Action items — with owners named, where the transcript makes them clear
- Decisions — when the recording contains real ones
One person thinking out loud — a brainstorm, a dictated draft, a running list:
- Key points — grouped by theme rather than in the order you said them
- Action items or Open questions — only when there genuinely are some
Here it organizes your thinking and keeps your voice. It won't turn a brainstorm into meeting minutes.
It judges this from what was said, not from who the microphone thinks was speaking — one mic in a room full of people records everyone as you, so the labels can't be trusted for this. Sections that would be empty or obvious are dropped rather than padded, and it won't add anything that isn't in your note or transcript.
Auto-Enhance
Auto-enhance after recording is on by default. Stop a recording and Enhance runs on it right away, showing you the changes to review.
To change it, go to Settings → Preferences and toggle Auto-enhance after recording.
This setting lives on the device, not your account, so each computer gets its own answer. On your phone, tap Enhance on the note when you've finished recording.
If the recording captured no speech, nothing runs, and you'll see "No speech was transcribed — nothing to enhance. You can still enhance manually."
Running It Yourself
On one recording. Open the transcript panel, find the recording in the recordings list, and click Add this recording to the note. This is the same path auto-enhance takes — use it when a note has several recordings and you only want to fold in the latest, or when auto-enhance is off. Recordings already folded in are marked with a check.
A recording that hasn't been transcribed yet can't be enhanced — you'll get "This recording has no transcript yet — nothing to enhance from it." Wait for transcription to finish and try again.
On the whole note. Click the chevron next to the wand button at the bottom of the note and pick Enhance. This reads the note and everything attached to it, and adds a section to the end.
The wand button itself runs Cleanup, not Enhance, unless you've changed your default — Enhance is one click further, in the menu. To make Enhance the default, clone it and tick Set as default sparkle target. See Skills.
The First Enhance Replaces the Note
When Enhance runs on a recording — the automatic path, and the per-recording button — what it does depends on whether the note has been enhanced before.
| What it does | |
|---|---|
| First time on a note | Compiles your notes and the recording into one clean document, replacing the note body. |
| Every time after | Adds a new section for that recording, leaving earlier ones alone. |
That's deliberate: the first pass turns scratch into a note, and later passes add to it without regenerating what's already there.
The first Enhance rewrites the whole note, not just the transcript part. It's instructed to carry your own writing through and keep every meaningful point you made — but it is a rewrite, not a merge. Read the diff before accepting.
You can override this per run: the diff has Append and Replace controls, and the chevron menu lets you pick a mode before running.
Reviewing and Undoing
Enhance never writes to your note on its own. The result is staged as a diff at the bottom of the note, and you choose:
- Accept — apply it
- Refine — give a follow-up instruction and run it again
- Reject — discard it
After accepting you get a toast — Note replaced or New section added — with an Undo button that puts the note back exactly as it was.
Undo goes back one step, and it isn't available on very large notes. If you're about to replace a note you care about, the safest move is reading the diff first, not relying on Undo afterwards.
What It Knows About the Recording
When Enhance runs on a recording, it's told more than just the words. Alongside the transcript it gets the note's title, whether the audio came from your microphone, system audio, or both, and roughly how many distinct voices were heard. If the note is linked to a calendar event, it also gets the event title and attendee count — which is how action items end up attached to the right names.
None of this is required. Enhance works on any recording, linked or not.