Data sovereignty that is a fact, not a setting
JesRecap records and transcribes on the workstation itself. After a one-time model download it needs no network at all — which makes “where is our data processed” a question with a one-word answer.
The public-sector objection to cloud notetakers
Usually it is not the transcription quality. It is jurisdiction, procurement and the audit trail.
Foreign cloud processing
Citizen and staff conversations processed on infrastructure in another jurisdiction raise transfer questions that can take longer to resolve than the project itself. JesRecap does not process meeting audio off the device, so there is no transfer to assess.
Bots in official meetings
An unaccountable extra participant in a council session, a case conference or a supplier negotiation is a governance problem before it is a technical one. Nothing joins your meeting.
Subscriptions and renewal risk
Per-seat monthly pricing means an annual renewal, a budget line that grows, and a dependency on a vendor staying in business. A one-time licence is a purchase, not a relationship you have to keep funding.
How JesRecap fits public-sector work
Air-gap friendly
The speech model is downloaded once from models.jesrecap.com; the speaker-label model is a one-time 54 MB download. After that, recording and transcription run with no network connection at all. A machine can be taken offline and keep working.
No foreign processing
Meeting audio, transcripts and metadata are written to the local disk as WAV files and a SQLite database. Nothing is sent to us, and there is no cloud region to choose because there is no cloud step.
Two tracks, retained
Microphone and system audio are captured separately and mixed only for playback. Because both originals are kept, a meeting can be transcribed again later — useful when a record is queried long after the fact.
Open formats for records management
Markdown, plain text, JSON, SRT/VTT captions and WAV audio. Everything you produce can be filed into your existing records system without a connector, an API key or a vendor-specific archive format.
Procurement-friendly licensing
From €34.95 once, per person. No accounts to provision, no seats to reconcile, no auto-renewal. Volume and departmental licensing is available on request through the enterprise page.
Small attack surface
No accounts, no telemetry, no integrations, no calendar access, and no cookies on this website. The two network calls JesRecap can make are a model download and licence activation or checkout.
What stays local
All meeting content: the two audio tracks, the mixed playback file, the transcript, speaker labels and the library metadata. Speech recognition runs on the workstation’s own CPU or GPU using a Whisper-family model. The only traffic is the one-time model download and licence activation or checkout — both of which can be done once, on a staging machine, before deployment.
No external AI service is involved at any point. By default, Meeting actions copy a prompt plus the transcript to the clipboard for use in whatever AI service your organisation has itself approved — an explicit human step, never an automatic upload. The optional AI notes on this device setting, off unless enabled, instead runs a language model on the workstation, so a summary can be produced without any external service at all.
Stated precisely
JesRecap holds no public-sector accreditation and we do not claim one. It is not FedRAMP authorised, not on a government cloud framework, and not certified against any national security standard — there is no cloud service to accredit, and desktop software cannot certify your process for you.
Because processing is on-device, JesRecap is designed to support your compliance process where cloud transcription is hard to approve: no third-party processor for meeting content, no cross-border transfer, and no vendor-held retention. The classification of a recording, its retention period, disk encryption, endpoint hardening and access control all stay under your control — which is the point.
Record with consent
Announce recording to participants and follow the recording, minute-taking and freedom of information rules that apply to your body. Local capture changes where the data goes, not what you owe the people in the room.
Questions
Can it run on a machine with no internet access?
Yes, once the speech model and the optional speaker-label model have been downloaded. After that, recording and transcription need no network. Licence activation is a one-time step you can do while the machine is still connected.
Is any meeting content processed abroad?
No meeting content is processed anywhere except the device. There is no cloud transcription step, so there is no region to configure and no transfer to document.
How does licensing work for a department?
Licences are one-time, by device count: €34.95 for 1 device, €49.95 for 2, €74.95 for 4, and €99.95 for 8 — no subscription. For volume or departmental licensing, contact us via the enterprise page — pricing is quoted rather than published.
Does anything join our meetings?
No. JesRecap records the audio the workstation already plays, using Core Audio process taps on macOS or WASAPI loopback on Windows, so the participant list is unchanged.
What about Linux?
Not today. JesRecap runs on macOS 14.2+ and Windows 10/11. There is no Linux build yet.
Related use cases
Same architecture, different room. Or see all industries.
Sovereignty by construction
macOS 14.2+ and Windows 10/11, offline after the one-time model download. From €34.95 once, no subscription. The 48-hour free trial is coming soon.