NoteCapt A2 Pro
A pocket-sized offline terminal for photos, audio and paper — paired with on-device AI that files everything the moment it lands.

- Cold start to capture
- 0.5s
- Microphone range
- 15m directional
- Capture modes
- Photo · Audio · Scan
- On-device AI
- Tag + file, local
- Network required
- None
- Storage
- Local only
Final specs pending hardware sign-off.

Four things happen without you doing anything.
Press once. It's already recorded.
0.5s from pocket to capture. No unlock, no app, no waiting for autofocus. Photo, 15m directional audio and paper scan on the same body.
- 0.5s cold start to first frame
- 15m directional microphone
- Paper and slide scanning
- Works with no signal at all

The photo lands on the audio timeline.
Whatever was being said the moment you took the shot stays attached to the shot. You find things by how you got them, not by filename.
- Photos pinned to the second they were taken
- Scan and audio share one timeline
- Searchable by moment, not by name
On-device AI tags and files it.
Before you put the device back in your pocket. The model labels, groups and files — it never rewrites what you captured.
- Runs entirely on-device
- No account, no upload
- Every tag traces to its source frame
Feed a finished folder to anything.
One action sends a digested folder to your local software or an external model — study cards, a recap, a paper outline.
- One-click export to external models
- Local-first, exportable any time
- Nothing locked to our platform
Two ways to keep it pointed at the room.
No case, no rig. It goes where your attention already is.

It snaps onto the back of a laptop lid and looks where the lid looks — at whoever is talking.

On a show floor there is no lid to clip to. The same device, worn, still reaches the person in front of you.
Rooms where the conversation is the thing you lose.
Trade show floor
Meeting room
Conference hall
Lecture theatre
Lab and group work
Site walkthrough
Inspection
It files what you captured. It does not rewrite it.
The on-device model does one job: label and file. It never paraphrases, never summarises into your archive, never invents a sentence you didn't record. Every tag traces back to the exact frame or second it came from.
- Labels, not summariesThe model tags and groups. It does not produce prose that goes into your archive.
- Every tag is traceableEach label points back to the exact frame or second it was derived from.
- Your words stay your wordsOriginal audio, photo and scan are never overwritten or paraphrased.
- Generation happens outsideWhen you want a summary or study cards, you feed a folder out on purpose.
Crowdfunding opens soon.
Early-bird units go to the list first.