Cloud meeting tools make leaving hard. Your transcripts live in someone else's database, behind an API you don't control, on terms that can change. Scriba is built the other way: every meeting you record lands in a SQLite database on your Mac first. Export is not an afterthought — it is how we prove you actually own what you captured.
A ZIP, not a vendor lock-in
When you export from Scriba, you get a standard ZIP archive (format version 1). Inside is a manifest.json describing what was exported, when, and whether audio is included, plus one JSON file per meeting with everything the app needs to reconstruct the record: metadata, transcript segments, summaries, knowledge items, notes, contacts, and chat history structures.
- manifest.json — format version, app version, export timestamp, meeting count, audio flag.
- Per-meeting JSON — transcripts, summaries, brains, templates, meeting notes, and linked contacts.
- Optional WAV audio — the same 16 kHz mono files Scriba plays back locally.
- Standalone contacts export — .scriba-contacts and .scriba-groups files for your address book.
No proprietary binary blob. No account required to read the archive on another machine. Unzip it, inspect the JSON, feed it into your own tooling if you want. That is portability in the literal sense.
Local-first, then sync
Authenticated users can mirror metadata and audio to AWS for cross-device access. Sync is convenient — it is not custody. Your Mac's database remains the source of truth. If sync pauses, you still have every segment. If you cancel a plan, your local files stay. Export is the escape hatch that works even when you are offline, on a flight, or moving to a machine that has never signed into getscriba.app.
We do not send a bot to your call. Scriba records from your Mac — mic, and optionally system audio via Screen Recording. Export gives you the same artifacts the app uses internally, not a watered-down summary page.
Import and merge, not just backup
Export is half the story. Scriba's import path reads the same ZIP structure, remaps contact IDs, and merges meetings into your existing library without clobbering unrelated work. Teams use it to consolidate laptops after a hardware refresh, to share a project archive with a colleague, or to keep a cold-storage copy on a NAS. Backups of the full user database use a sibling format with backup-manifest.json — same philosophy, broader scope.
What we don't hold hostage
Managed AI and transcription are services on top of your data, not a cage around it. Pro and Max users get AssemblyAI streaming and batch transcription; Free and BYOK users route through their own keys. None of that changes what sits on disk. If you outgrow Scriba, you leave with searchable transcripts, not a PDF and a prayer. That is the deal we think productivity software should make.