Built for privacy-first professionals
Record, transcribe, and summarise meetings, sessions, and dictation — fully on your iPhone, iPad, Mac, or Apple Watch. No meeting bot. No third-party server. No compromise.
Most note-taking apps send your audio to their cloud the moment you tap record. LocalScribe doesn't. Transcription, speaker detection, and summaries all run on your device by default. If you ever want cloud accuracy, it's opt-in — per recording.
On-device by default
Recording is real-time. Transcription is real-time. Summaries take a few seconds on the device itself. Nothing about your audio touches a third-party server unless you explicitly opt in.
When you want cloud accuracy
Optional cloud features upgrade summary quality (GPT-4o or o3) and speaker naming (Claude Opus 4.7 — the model with the lowest published hallucination rate). Audio stays on your device. Only the transcript text is sent.
Wherever you work
Start a recording on your Watch from across the room. Review the transcript on your laptop. Read the summary on your phone. Every device uses your iCloud account — no LocalScribe account, no third-party sync server.
Built for serious work
Every recording becomes a study guide, a fact-checker, and a reference library — without ever sending your audio anywhere. New capabilities for students, professionals, and anyone who reads what they record.
When the speaker says "this is on the test," "remember this," or "key point" — we pull it out and put it in a callout at the top of your summary. So the stuff you actually need to know is the first thing you see.
Tap Audit and the AI re-reads your summary against the original transcript, flagging factual errors, contradictions, wrong-speaker attribution, and invented specifics. Underlined inline. Tap any flag to see exactly what's off and why.
Set your profession to Education and every recording surfaces academic papers (OpenAlex) and educational videos (YouTube) matched to the topics in your lecture. Free sources, abstracts included, one tap to the PDF or video.
Ask a natural-language question — "what did Sarah think about pricing," "anything related to legal review" — and we surface the moments that match, ranked by meaning, not just keywords. Filter to one speaker's words to read just their side of every conversation.
For therapy, sales calls, and hard conversations — multimodal AI listens to the actual audio plus reads the transcript, then writes a structured read of tone, hesitation, emphasis, and emotional shifts. Quotes the exact words that carried weight.
Two voices merged into one cluster? Long-press to reassign, or multi-select and split into a new speaker. Every fix teaches the per-user voiceprint store, so the next recording with that person gets it right automatically.
If your job depends on confidentiality, LocalScribe was designed for you.
Attorneys
Depositions, client interviews, witness prep, legal dictation. Specialised summary formats.
Clinicians & therapists
Session notes, patient encounters. Audio stays on device — HIPAA-friendly posture by default.
Journalists
Source interviews, conference recordings. Protect your sources — nothing leaves your device.
Students & researchers
Lectures, focus groups, qualitative interviews. Auto-formatted in the style your field uses.
Executives & founders
Board meetings, 1-on-1s, strategy sessions. Notes shaped to each meeting type.
Advisors
Real estate, financial, planning. Client meetings stay compliant and private.
Free for casual use. Premium tiers for when you want the best.
Try LocalScribe with everything on-device.
Unlimited recordings, with the best AI on demand.
For heavy users and the highest-stakes work.
Yes. By default, audio is transcribed and summarised on your device's Neural Engine — nothing leaves. Premium users can opt-in to cloud features for higher accuracy, but it's per-recording and clearly labelled. A "No Cloud" toggle pins sensitive recordings to fully on-device.
No. LocalScribe records directly from your phone, laptop, or watch microphone. That means it works for in-person meetings, phone calls, dictation, lectures, depositions — anywhere there's sound. No Zoom integration required.
iPhone, iPad, Mac (Apple Silicon, via Mac Catalyst), and Apple Watch. Recordings sync through your iCloud — start a recording on your watch, finish reviewing on your laptop.
Only when you opt-in to a Premium cloud feature for a specific recording — and only the text (not audio) goes to OpenAI for summaries and speaker reasoning. The Whisper transcription path is separately opt-in and labelled. Both providers contractually do not train on customer data.
Every LocalScribe recording shows its exact processing path: which engine transcribed it, whether cloud features were used, whether audio left the device. That metadata is exportable. The "No Cloud" override is per-recording and persisted, giving you a verifiable compliance trail.