On the record.Off the cloud.

Pick the trade-off each recording deserves: Apple, Parakeet or Whisper. Choose the language first and Kayit grays out any engine that can't handle it, with the reason and a time estimate, before you commit. All three run on-device; none of them is a network call after the initial download of the engine.

Edit paragraphs with real Mac text editing, and the timestamps survive. Fix a typo and every word keeps its exact timing, so playback stays aligned to the words you corrected. Autocorrect, smart quotes and Writing Tools are switched off in the editor, so nothing can rewrite a source's words for you.

Share straight from Voice Memos, which has no export of its own. Right-click ▸ Open With in Finder, or drop a file on the window. On iPhone and iPad it is the share sheet, from Voice Memos or Files. Drop in a video and Kayit takes the audio track, leaving your file untouched.

Automatic speaker separation splits the transcript every time the speaker changes and labels who is talking. Rename a speaker once and it applies across the whole interview. Save a sample of your own voice from a transcript you have already labeled, and Kayit names you automatically in every interview after that.

An on-device model flags admissions, allegations, contradictions, on-record claims, and financial and legal exposure, linking every one back to the moment in the audio. Set each category's own bar, and see roughly how many passages that will flag before you re-run. Click a passage to see it in context in the transcript. The passage is the transcript's own words rather than a paraphrase, with a line under it from the model saying why it was flagged.

A plain-language summary, a timestamped outline whose entries seek the audio, and suggested quotes anchored to the transcript's real words rather than a paraphrase. It runs on your Mac, and a switch turns it off, enforceable across a whole organization if your policy requires it.

A project is a folder in your library, so what you see in the sidebar is what the Finder and your backup see, and nothing moves until you ask. One roster of speaker names runs across the project, so a source you have interviewed six times is spelled the same way six times; every quote you saved in it gathers into a single list; and the whole set exports together, audio included. The running list of what you still need lives in the project as a plain Markdown file. Making a project is the part that needs Pro: if a subscription lapses, the projects you have stay open and you can keep filing into them, archiving them and finding what is inside them.

Export the transcript and the audio as one bundle. Someone without Kayit double-clicks the transcript and reads it; someone with it gets the speakers, word timings and your saved quotes without re-transcribing. Sending the words alone costs nothing: plain text is in the free tier, and Pro adds Markdown, CSV, SubRip subtitles, timestamps and anything that carries the audio.

One purchase covers all three. The iPhone and iPad app transcribes, plays back, edits, names speakers, saves quotes and exports, with the same free line drawn through it, and the iPad lays that out in three panes with the insights beside the transcript. The AI insights are the one thing the iPhone does not have, for the plain reason that the model needs more memory than a phone has. A saved sample of your voice works on each device but does not yet travel between them. None of the three records; all of them transcribe recordings you already have.