Skip to content

Siri Recap & Live Rewind vs. Ayumi

Apple Watch Series 12 and Apple Watch Ultra 4 add two Audio Intelligence features: Siri Recap and Live Rewind. Both turn speech into text on your wrist, which raises a fair question: where does Ayumi fit?

The short answer is that they solve a different problem. Siri Recap and Live Rewind are built to forget — they keep a short summary or a 15-second snippet and throw the audio away. Ayumi is built to keep — your own words, in full, as files you own.

Siri Recap takes ambient notes on the conversations around you, on a schedule or when you switch it on, and produces a short, high-level summary. It does not produce a transcript. Audio is processed in a protected buffer and deleted once processed, so there is no recording to go back to. Summaries appear in the Recaps tab of the Siri app and are deleted after 7 days unless you save or export them.

Live Rewind shows the last 15 seconds of speech as text when you double-press the Digital Crown. It is a “what did they just say?” tool. An unsaved snippet disappears shortly after the display dims.

Neither feature attributes statements to speakers, and Apple notes that both can be incomplete or inaccurate.

Ayumi is a voice journal. You decide to record, you speak, and the recording becomes a journal entry:

  • The full transcript is kept. On-device transcription writes a verbatim transcript; nothing is condensed away unless you choose a profile that summarizes.
  • The audio is kept. The recording stays attached to the entry, so you can listen again or re-transcribe it later with a different engine.
  • Everything is a file you own. Each entry is a Markdown file in a folder you choose, with no expiry and no account. See Concept.
Siri RecapLive RewindAyumi
PurposeRemember the gist of conversationsCatch a line you just missedCapture your own thinking, in full
What it listens toConversations around youThe last 15 seconds of speechWhat you record, while you record
How it startsA schedule (time or place), or switched on manuallyDouble-press the Digital CrownTap Record, Siri, a complication, the Action Button, or a Shortcut
OutputTitle and key points — no transcriptA 15-second text snippetVerbatim transcript with a title, or whatever your profile asks for
Original audioDeleted after processingNot keptKept, attached to the entry
Retention7 days unless saved or exportedGone after the display dims unless savedUntil you delete it
Where it livesSiri app, synced through iCloud with end-to-end encryptionSiri app, if savedMarkdown files in your own folder or iCloud Drive
ProcessingTranscribed and condensed on iPhone, summarized by Private Cloud ComputeOn Apple Watch and iPhone; Private Cloud Compute if you ask Siri about itOn-device by default; a cloud engine only with your own API key and consent
Speaker labelsNoNoAvailable with Gemini Transcribe, Muse Voice Transcribe, and xAI Grok
Can you edit or reuse it?Edit in the Siri app, export to Notes or JournalSave or ask SiriPlain Markdown — open in Obsidian, an editor, or any AI tool
DevicesSeries 12 or Ultra 4, with iPhone 16 or laterSeries 12 or Ultra 4, with iPhone 16 or lateriPhone (iOS 26 or later) and Mac (macOS 26 or later); Apple Watch on watchOS 10 or later

The difference that matters: summary vs. your words

Section titled “The difference that matters: summary vs. your words”

A summary is someone else’s reading of what you said. It’s useful for a conversation you were part of, where the point is to recall the outcome. It is much less useful for your own thinking, where the half-formed sentence, the tangent, or the exact phrasing is often the part worth keeping.

Siri Recap is deliberately designed so that the original is gone: no audio and no transcript, only a condensed version. That is a strong privacy choice for a feature that listens ambiently to other people. But it also means a recap can’t be checked against what was really said, and it can’t be re-read differently later.

Ayumi takes the opposite position for a narrower job. It only listens when you ask it to, and when it does, it keeps everything — so that a year from now you, or the AI tool you point at your journal, can still work from the original.

Both approaches take privacy seriously, but they protect different things.

  • Siri Recap protects by not keeping. Raw audio is never stored, transcripts are never produced, summaries filter out sensitive details such as financial data and identifiers, and unsaved recaps expire after 7 days.
  • Ayumi protects by keeping it with you. Recordings and transcripts stay on your device or in your own iCloud Drive folder. Ayumi has no server and no account. Transcription is on-device by default, and audio leaves the device only when you choose a cloud engine with your own key — after a consent prompt.
SituationBetter fit
You missed what someone just saidLive Rewind
You want the gist of a day of conversations without recording anythingSiri Recap
You’re thinking out loud on a walk and want every wordAyumi
You want a voice journal you can reread, search, and keep for yearsAyumi
You want to feed your own words to an AI tool laterAyumi — plain Markdown, full transcript
You need to check exactly what was saidAyumi — the audio is still there

They also work together. A recap exported to Journal or Notes reminds you what happened; a short Ayumi recording afterwards captures what you think about it.

Ayumi’s watch app works on any Apple Watch running watchOS 10 or later, paired with an iPhone running Ayumi — no Series 12 required. Record on your wrist and your iPhone transcribes it into your journal. See Apple Watch and Voice Recording.

Download Ayumi on the App Store

From the same maker: Kitae, a dead-simple strength-training log

Follow Ayumi: X (@AyumiJournalApp)