JesRecap is a meeting recorder for Mac and Windows that records your microphone and the meeting audio, transcribes on your device, and never uploads a word. Today we are opening the preview release channel — version 0.1.6 — and we would rather tell you exactly what state it is in than pretend it is a 1.0.
Where it comes from
We make JesType, an offline dictation app that runs speech recognition entirely on your own machine. Building it taught us two things. First, on-device speech models are now genuinely good enough for real work. Second, the people who care most about that are the people whose employer will never approve a cloud transcription service — clinicians, lawyers, people in government and finance, anyone whose conversations are other people's private matters.
Those users kept asking the same follow-up question: can it do my meetings? JesRecap is the answer, built on the same principle. Pay once, runs on your device, no cloud.
What works today
Dual-track capture, with no bot
JesRecap records two separate tracks — your microphone, and the audio your computer plays out, which on a call is everyone else. macOS uses Core Audio process taps; Windows uses WASAPI loopback. Nothing joins the meeting, so there is no extra participant in the list and no consent-by-bot moment. It works with Zoom, Teams, Meet, Webex, Slack huddles, any conferencing app at all — and with in-person conversations through the microphone.
After the recording stops, the two tracks are mixed into a third file for easy playback. All three stay on your disk, so you can retranscribe a meeting later without recording it again.
Local transcription
You pick one Whisper-family model, it downloads once from
models.jesrecap.com, and after that transcription happens
offline against the local file. Live transcript during the meeting is
optional — turning it off keeps memory use lower and defers all model work to
the moment you press stop.
Speaker labels
A local diarization model (a one-time 54 MB download) separates your voice from the other participants after recording. Because your microphone is already its own track, "You" is exact rather than inferred. Other voices are labelled Speaker 1, Speaker 2, and so on — the app does not know their names and will not guess. How speaker labels work →
Meeting actions
There is no cloud AI in JesRecap, on purpose. One-click actions copy a ready-made prompt with your transcript already included — summary, action items, or your own custom prompt — and you paste it into whichever assistant you already use and already trust. The prompt templates are editable in settings. Nothing is sent anywhere by the app. See the actions →
And if you would rather not involve an assistant at all, there is now a setting called AI notes on this device — off unless you turn it on — that has those same buttons write the note locally, using a language model that downloads once (about 1.3 GB) and then runs on your own hardware. The note appears under the buttons, is saved with the meeting, and can be cancelled mid-generation. Still nothing over the network.
Exports
Markdown notes, plain text, a JSON archive, SRT and VTT captions, and the WAV audio — mixed, microphone-only, or system-audio-only. Ordinary formats, so your meetings remain readable by tools that are not ours.
A searchable library
Meetings are listed in the sidebar and searchable by content. Metadata lives in a local SQLite database next to the audio files, under your application support folder. If you used JesScribe, your existing meeting data is migrated automatically on first launch.
What is not done yet
This is a preview, and the honest list matters more than the feature list.
- The 48-hour free trial is not live. The trial flow exists in the app, but email delivery of trial codes is not switched on yet, so it is marked "coming soon" everywhere on this site. When it works, we will say so.
- Polish. Long meetings, unusual audio device setups, and mid-recording device changes are the areas we are still hardening. Error states could explain themselves better than they currently do.
- No Linux build. System-output capture is implemented for macOS and Windows only today.
- No integrations. No calendar sync, no plugins, no team or cloud features. Some of that we may build; some of it we will not, because it would mean putting your meetings somewhere other than your machine.
We would rather ship a small, honest app and grow it than announce a roadmap we have not written yet.
Price
€34.95, once. No subscription, no per-minute transcription charge, no seat count. One license covers 1 to 8 devices (by tier) belonging to the same person. There is no account to create — activation is a license key, and you can manage or recover it from this site.
A one-time price is the natural fit for local software. There is no server transcribing your audio, so there is no recurring cost for us to pass on to you.
Try it, and tell us what breaks
JesRecap needs macOS 14.2 or newer (Apple Silicon) or Windows 10/11. On macOS it uses the system Audio Capture permission, which you grant once and can revoke at any time.
If something goes wrong — a track that came out silent, a transcript that read badly, a device the app did not recognise — write to support@jesrecap.com. Preview feedback is what decides what 0.2 looks like.