Recap: On-Device Meeting AI

Recap records your meeting and keeps the audio. Not a summary of what was said — the full transcript, and the recording it came from, so you can check any line yourself. Recording continues while the screen is locked or you are using another app, because a meeting does not stop when you put your phone down. A phone call pauses the recording rather than ending it, and the same file resumes when the call is over. WHAT YOU GET - Full transcript, not a summary - The original audio, kept on your device - Tap a line to jump to that moment in the recording - Lines with lower recognition confidence are marked, so you know what to check - Titles, times and attendees filled in from your calendar, if you allow it - Export as PDF, plain text, or a single bundle with the audio included - No length limit — a three hour recording is transcribed in one pass WHAT YOU DO NOT GET - No subscription - No free trial that turns into a charge - No account, no sign-in, no email address - No upload of your recordings or transcripts HOW IT WORKS Speech recognition runs on your device using Apple's on-device speech engine. Your recordings and transcripts are not uploaded and are not sent to a server. They stay in the app until you delete them or export them yourself. The app is free and supported by a banner ad, which is the only part of the app that connects to the network. WHO IT IS FOR People who need to be able to check what was actually said — not just a rough account of it. Journalists, researchers, consultants, anyone who takes notes in a meeting and would rather have the recording to fall back on. MORE ABOUT THIS APP Full feature list, screenshots and answers to common questions: https://todays-tasks.com/apps/recap-meeting-recorder/ Recap is one of a set of small, privacy-first tools built by an independent developer in Hong Kong. The rest of them, including twelve browser tools that run without uploading anything, are at: https://todays-tasks.com REQUIREMENTS iOS 26 or later. On-device speech recognition requires the language model to be downloaded, which iOS handles the first time you transcribe.