Earkeep
Decide on evidence, not opinion, by keeping every session word for word
The moderated session, the debrief right after it, the synthesis that turns both into a claim: research is only as good as the fidelity of what you kept, and it is delivered as themes, a report and a repository that has to outlive the study. Earkeep is an always-on transcriber that keeps every session on your own device, with nothing new appearing on the table and no bot in the call.
No bot · No cloud · No record button
14-day full trial, no account required.
The sessions that decide it
Moderated session
What has to survive it is a clean, complete transcript with timestamps, because a highlight reel and a persona are only ever as traceable as the words underneath them. The instrument itself is part of the problem: a meta-analysis of 28 studies covering 2,586 participants found 48% of people noticing a camera and 15% reporting that it changed how they behaved.
With Earkeep: nothing new appears on the table and nothing joins the call, because capture is already running and there is no record button to press in front of the participant. Audio is transcribed on your device and never written to disk, so the session never reaches a vendor. How private-by-design works →
Note-taker debrief
What has to survive it is the first-pass impression, in the ten to fifteen minutes before the next session starts and the two blur together. That window is exactly when a cloud tool is still uploading and has nothing to give you.
With Earkeep: transcription happens locally as the day goes, so the debrief works from the real transcript, not from memory. The session, the debrief and the follow-up an hour later are one continuous timeline, and you mark each as its own span. How always-on works →
Synthesis workshop
What has to survive it is the theme with the verbatim quotes still attached, because a finding a stakeholder cannot trace is a finding they can wave away. dscout puts synthesis at two to three hours per hour of interview, and notes that carving out thirty hours for ten hours of interviews is out of reach for most teams.
With Earkeep: every transcript is a plain text file on disk, so you can point an agent at a whole month of sessions and ask what recurred, where two participants contradicted each other, or for every verbatim mention of one workflow. It is a reviewed first pass into that ratio, not a verdict. How agents work in Earkeep →
The rest of the week
Those three are where the evidence is made. They are not where the week goes: a study round is three to seven interviews and two to five usability sessions on top, each with its own debrief. Articos puts the same ratio at three hours of analysis per hour collected, which worked out to roughly eighteen hours of synthesis for eight 45-minute sessions.
| Meeting | You | What has to survive it |
|---|---|---|
| Usability test session | Run | Task success and failure moments, verbatim reactions, timestamps for the clips |
| Stakeholder readout | Run | Findings turned into a recommendation somebody will actually act on |
| Intake and scoping call | Run or attend | An agreed research question and success criteria, before recruiting starts |
| Recruitment or screener review | Attend | A confirmed, scheduled, consented participant list |
| Research team or ops sync | Attend | Shared status, so two researchers do not run the same study twice |
| Cross-functional critique | Attend | Evidence injected into a decision that is already in motion |
| Diary study or field check-in | Run | Longitudinal notes tied back to the study timeline |
Earkeep does not ask you to choose between them. Every scheduled row above starts, ends and titles its own span from your calendar, so a seven-session week ends as named transcripts with no export step and nothing left to file.
Templates are the one thing this list mostly does not need, and it is worth saying so: a moderated session already has a script and a protocol, and a generic minutes template would be a downgrade. The exception is the stakeholder readout, which is a meeting like any other, and the shape it wants is on the meeting minutes hub.
The craft the sessions themselves run on has three pages of its own, written to be read together: open-ended questions on form against substance and why the funnel is mandated in one profession, establishing rapport on the three components and the evidence that it yields information, and empathic listening on what to do with an answer once it arrives. Each ends with what you can check by reading your own session back.
See it in action
A day of sessions, end to end
Every moment sits on a scrollable timeline, so a session, the debrief after it and the aside an hour later are all findable in one place. You scrub to the moment a participant hesitated and read exactly what they said.
Synthesis as a reviewed first pass
Point the agent you already use at a month of sessions and ask what recurred, where participants disagreed, or for every quote about one task. It reads the transcripts, and you review a draft with the verbatim quotes attached, which is the part that makes it checkable rather than authoritative.
Pricing: one payment, no subscription
€39
once, for one device, not a subscription.
No tiers, no per-seat billing, no recurring bill. Try it free for 14 days with full functionality, no email and no account required. See full pricing and policy details →
FAQ for UX researchers
Does Earkeep record participants without their knowledge?
No. Earkeep is always listening on your device while it runs, but it does not join calls, announce itself to other participants, or replace your obligation to obtain informed consent. Recording a research participant, local processing or not, still requires telling them and getting their agreement, exactly as it would with any other recorder or platform transcript. NN/g's guidance also expects a valid consent form to say when the raw recorded data will be deleted, if ever, which is a question you answer, not Earkeep.
Where does a participant's audio actually go?
Nowhere. Audio is processed only in memory on your device and is never written to disk, never uploaded, and never sent to a transcription API. Only the text transcript is saved locally, as a plain file you control.
Can I delete one participant's session without touching anything else?
Yes. Transcripts are plain, human-readable files on your disk, organized by day and by the spans you or your calendar mark. You can open, edit, or delete any session's data directly, the same way you'd manage any file.
Does this replace my research repository, like Dovetail or Condens?
No, and it isn't trying to. Earkeep is the capture and transcription layer. It gets a clean, complete transcript of every session without a manual tagging bottleneck. You still push synthesized findings into whatever repository your team uses.
Does Earkeep support the languages my participants speak?
Whisper covers 99+ languages; Parakeet-TDT v3 covers 25 European languages with faster, more accurate English transcription. See the full supported-language list for both models.
Will always-on capture slow down my device during back-to-back sessions?
No more than a normal transcription app. A voice-activity-detection pipeline only runs transcription when someone is actually speaking, so silence between sessions costs nothing extra.
Stop losing what participants said
14-day full trial, no account required.