Earkeep for UX researchers

A UX researcher's job isn't running sessions. It's turning what users actually said into evidence a product team will act on, fast enough that the decision hasn't already been made without them. Every section below comes back to that: capture that doesn't change how participants behave, transcripts the moment a session ends, and synthesis time given back to actual analysis.

Researchers can run the interviews. It's everything after that piles up: "i can run about 6 or 7 a week no problem. it's the transcribing, tagging, pulling themes, then turning all of it into something a PM will actually read instead of skim is what i'm finding difficult… once we crossed maybe 30 sessions a month the manual coding just fell apart."

That's a familiar week: three to seven interviews or usability sessions, debriefs right after each one, stakeholder kickoffs and readouts, team syncs, and for consultants, workshops and field studies on top. One researcher described wrapping 10 interviews in two days and then "manually tagging transcripts in Google Docs and it's pretty painful." A research lead put a number on what gets lost: one researcher on the team spends 6 to 8 hours a week manually tagging and coding transcripts, while feedback from Slack, internal meetings, and tickets "just doesn't get captured at all," leaving the team working from maybe 15 percent of actual customer voice.

Two more pains sit underneath the time problem.

A recorder on the table changes the room. "the room does change a bit when there's a phone or recorder sitting there. People get a little more careful. I get more aware that I'm 'recording a meeting' instead of just having a conversation." That's a hit to the very data the session exists to collect.

Transcripts alone lose signal. "A participant might say they like a feature but when you watch the recording you notice a long pause before the answer… None of that really shows up in the transcript." Researchers end up going back to the source anyway, which means what's kept has to be findable and safe to keep.

Earkeep listens continuously to mic and system audio, without a start button and without a bot joining the call. There is nothing on the table to make a participant more careful, and nothing to forget to start before the session goes ahead.

• Nothing to configure per session. The interview, the debrief you do right after it, and the hallway follow-up an hour later are all already in the day's transcript. You mark the span that was the actual research session after the fact, on a timeline.
• Calendar auto-spans for scheduled sessions. Connect Google Calendar, Apple Calendar, Outlook, or any ICS calendar URL, and a scheduled user interview starts and ends its own span automatically.
• Transcribed by the time the session ends. Whisper or Parakeet models run locally and transcribe as the audio comes in, so there's no transcription queue to wait on before synthesis can start.

The 6 to 8 hours a week of manual tagging becomes pointing an agent at a month of sessions and asking for themes, direct quotes, and contradictions, while staying in the loop on what it surfaces rather than trusting it blind.

• Capture: Otter, Rev, or the transcript built into Zoom or Meet. See Earkeep vs Otter for the capture-layer comparison: bot in the call and audio on a third-party server, versus nothing on the table and nothing that leaves your device.
• Local, manual transcription tools like MacWhisper transcribe locally too, but start and stop per file or per session, one recording at a time. See Earkeep vs MacWhisper.
• Repositories and manual tagging: Dovetail, Condens, EnjoyHQ, Marvin, Looppanel, and BuildBetter, with Google Docs and rainbow-sheet spreadsheets doing the tagging NVivo or Atlas.ti would do more rigorously. Earkeep isn't a repository; it's the capture and transcription layer feeding one.
• Ad-hoc synthesis: NotebookLM, Claude, and Gemini, used daily but with real skepticism. As one researcher put it, "If I took the first pass of AI analysis as gospel, I would routinely end up working from false or distorted findings." Earkeep's in-app agents work the same way: a first pass you review, not a verdict you accept.

€39, once, for one device. No subscription, no per-seat pricing, no tiers, and no recurring line item on a research budget.

• 14-day free trial, full functionality, no email address and no card required to start.
• Yours to keep, forever. Includes all updates to Earkeep 1.x. If a paid major upgrade ever ships, the version you bought keeps working.
• 14-day money-back guarantee, no questions asked, if you buy and it doesn't fit your workflow.
• No account, no cloud, no API keys, ever.

See full pricing details →