Privileged conversations stay on your own disk
Client calls, intake interviews, depositions and internal case discussions — captured as two separate audio tracks and transcribed locally. No bot joins the call, and no third party ever holds the recording.
Privilege and cloud transcription sit badly together
The risk is not that the vendor is careless. It is that the recording exists somewhere you cannot speak for.
A copy outside the firm
Once privileged audio is uploaded, a copy of the client’s words lives on infrastructure your engagement letter never mentioned. Explaining that to a client — or to opposing counsel — is not a conversation anyone wants.
An extra attendee on the record
Notetaker bots show up in the participant list of a call. In a negotiation or a witness interview, an unexplained attendee is a distraction at best and an objection at worst.
Summaries instead of a record
Tools that only keep an AI summary throw away the thing you actually needed. For case work the source matters: what was said, in what order, in whose voice.
How JesRecap fits legal work
Dual-track capture
Your microphone and the other side of the call are written to separate WAV files — microphone.wav and system-audio.wav — and mixed into a playback track only afterwards. Two tracks means one side’s crosstalk never buries the other’s answer.
Depositions and in-person intake
In a room, the microphone track is the whole story: put the laptop on the table and record. On a call, JesRecap takes the audio your computer already plays via Core Audio process taps on macOS or WASAPI loopback on Windows — Zoom, Teams, Webex, Meet, a phone bridge on speaker, anything.
Speaker labels
A local diarization model separates “You” from other speakers after the recording, which makes a long interview readable. Treat labels as a drafting aid and confirm them yourself before a transcript is relied on.
Retranscribe later
Both original tracks are kept, so any meeting can be transcribed again — with a larger model, or after you notice the first pass mangled a name. The audio is the record; the transcript is a rendering of it that you can redo.
Exports for the case file
Markdown notes and plain text for the matter folder, a JSON archive when you want the structured version, SRT/VTT captions for video evidence, and the WAV audio itself — mixed, microphone-only or system-audio-only.
A searchable library
Every recording lands in a searchable local library, backed by WAV files and a SQLite database in your own user folder. No matter-management integration required, and nothing to sync.
What stays local
Audio, transcripts and metadata never leave the machine. Transcription runs on your own hardware with a Whisper-family model downloaded once from models.jesrecap.com. The only network traffic is those one-time model downloads and licence activation or checkout — no accounts, no telemetry, no meeting content.
There is no cloud AI either. Meeting actions copy a prepared prompt with the transcript to your clipboard so you can paste it into whatever AI tool your firm permits — a decision that stays with your firm, not with us. Or turn on AI notes on this device, off by default, and the summary is drafted by a model on your own machine: no disclosure to a third party, and nothing to waive privilege over.
Careful wording, on purpose
JesRecap is a recorder and a local transcriber. It is not a certified evidence-handling system, it does not apply a chain-of-custody seal, and it makes no representation about the admissibility of anything it produces. Whether a recording is lawful, privileged, disclosable or usable in a given forum is your judgement to make.
What we can state precisely is architecture: on-device capture, on-device transcription, original tracks retained, no upload. That is what supports your own confidentiality and record-keeping process — the process stays yours.
Record with consent
Recording rules vary by jurisdiction and by forum, and some require every participant’s agreement. Say clearly that you are recording, get consent on the record where that applies, and follow your bar or law-society guidance.
Questions
Does the recording leave my machine at any point?
No. Capture, storage and transcription are local. JesRecap's only outbound requests are the one-time speech model download and licence activation or checkout.
Why two audio tracks instead of one file?
Because a mixed file loses information. Keeping microphone and system audio separate preserves each side of the conversation, lets you export one side alone, and makes a later retranscription cleaner. The mixed playback track is generated after the meeting for convenience.
Can I transcribe an old recording again?
Yes. Both original tracks stay on disk, so you can rerun transcription on any meeting in the library later.
Will a bot appear in a client call?
No. JesRecap records the audio your computer already plays rather than joining the meeting, so the participant list is untouched. Announce the recording anyway.
Does it summarise the call for me?
Not itself. It copies a ready-made summary or action-items prompt together with the transcript to your clipboard, and you paste that into whichever AI your firm has approved. Nothing is sent anywhere by JesRecap.
Related use cases
Same architecture, different room. Or see all industries.
Keep privileged audio where privilege lives
macOS 14.2+ and Windows 10/11. from €34.95 once, no subscription. The 48-hour free trial is coming soon.