Talk like a human. Get back structured tasks.

Built for the moments typing is friction

Walking to the car. Mid-conversation. In the shower with your phone on speaker. The morning brain dump before the day starts. Driving (legally, hands-free). The most useful thoughts you have don't show up at your laptop — they show up when your hands are full.

Voice capture is the surface for those moments. You don't need to remember a syntax. You don't need to format anything. You talk like a person, and the app does the parsing.

How it works

1. Press Shift + N on desktop, or tap the microphone icon at the top of the screen on mobile.
2. Talk. Say everything you want to capture. When you stop talking, the app picks up that you're done.
3. The transcript appears in a review sheet — the full thing, exactly as you said it.
4. Below the transcript, AI's task extraction: each candidate to-do shown as its own row, with chips for area, project, tags, and due date already filled in where the model could tell.
5. Edit anything. Uncheck anything you don't want saved. Hit Save. Selected items become real tasks; the original transcript stays around as a note for traceability.

If you said "I should email Jane about the contract, and I'm thinking about how to structure the Q3 OKRs," you'll see two task candidates ("Email Jane about the contract," "Structure Q3 OKRs") and a saved note containing the full transcript.

Reflective vs structured — automatically routed

Not everything you say is a to-do. Some of it is thinking out loud. Mind Like Water classifies each chunk and routes accordingly:

• Discrete tasks — concrete actions ("Email Jane," "Pick up cake Friday") → review sheet, your call to save.
• Reflective content — first-person thinking, journaling, processing ("I'm worried about the renewal because…") → appended directly to today's Daily Page without interrupting you to "review" anything.

The distinction matters: a stream-of-consciousness brain dump shouldn't fill your Inbox with phantom tasks. It should land where journal content belongs. So that's where it goes.

What gets pulled out automatically

For each extracted task, the AI proposes the same metadata as AI capture — because it's the same engine running on the transcript:

• Area — matched against the Areas you've defined
• Project — matched against your active projects
• Tags — existing or worth-creating
• Due date — when there's a date phrase in what you said
• Status: Waiting — when you said you're blocked on someone

Voice gets a slightly looser confidence threshold than typing — spoken capture tends to be looser, so the extractor is more aggressive about catching what you meant.

What this gets right that other tools miss

You always see what you said before it saves

The full transcript shows up at the top of the review sheet. If something got mangled (rare, but it happens with proper nouns or strong accents), you can edit the task title inline before saving. The underlying transcript stays exactly what you said — searchable, editable, yours.

Brain dumps don't pollute your Inbox

We've used voice tools that took an eight-minute morning ramble and turned it into 47 phantom tasks in your inbox. Mind Like Water's reflective routing keeps that from happening. If you weren't actually committing to a thing, the model doesn't pretend you were.

Mobile-first because that's where voice happens

The microphone is right there in the top header on mobile. No menu, no setting to find. The phone is where most voice captures happen, so that's where the affordance is biggest.

Privacy

Voice capture sends audio to Whisper for transcription and the resulting transcript to Claude for extraction. Once the extraction completes, the audio is dropped. We don't keep recordings. The transcript is saved as a note in your account so you can audit what was captured; deleting the note deletes the transcript.

Pro accounts get a faster, higher-accuracy Whisper variant; free accounts use the standard one. Both are accurate enough for everyday speech.

What this doesn't do (yet)

• No on-device transcription. Both Whisper paths are cloud. On-device transcription is on the roadmap as part of Private Mode; until then, audio leaves your device.
• No streaming transcription. We transcribe after you stop speaking, not as you go. Faster overall, but no partial text mid-sentence.
• No voice commands. Talking doesn't trigger app actions other than "save what I just said." We'd rather not encourage rambling at the app.

Frequently asked