What's Coming
The next scroll.
Thoth 1.8.1 is live on the Mac App Store and 1.0 is live on iPhone and iPad. Here is what has shipped, what I am building next, and what you get to vote on.
What Thoth does today.
Everything below is live on the Mac App Store and, where it applies, on iPhone and iPad. Version by version detail is in the app, under Settings.
Recording, however you work
Quick actions in the menu bar. A nudge when Zoom, Teams or Slack launches, and an offer to record when a meeting-shaped event starts on your calendar. Mic, system audio, or both on separate channels, with no virtual driver and no bot joining the call. Folders three levels deep, colour-coded tags, and automatic AAC compression so the library stays small.
Four engines, ninety-nine languages
WhisperKit for all 99 languages, Parakeet TDT v3 for speed, Parakeet EOU streaming word by word as you speak, and Apple Speech on device. A floating live panel shows the text as it arrives. Speaker identification is aligned word by word rather than estimated. Feed it a custom vocabulary from a CSV. Smart editing highlights the words the engine was unsure about so you can fix them in place, or run an AI proofread over the whole transcript. How accuracy is measured →
AI that does not leak
Anonymization runs on your device: people, places, organizations, emails, phone numbers, card numbers and IDs validated by checksum, plus patterns you define yourself. You see every detected item as a token and decide what leaves. Summaries run on device through Apple Intelligence or five bundled local models. For cloud, bring your own key, use a verified hardware enclave whose attestation is checked before every session, or point Thoth at your own OpenAI-compatible server. How redaction works →
Everywhere, without a middleman
iCloud sync carries your whole library, recordings, transcripts, summaries and notes, across your Macs and your iPhone through your own iCloud account. Or send a recording straight from Mac to iPhone over AirDrop or local Wi-Fi, with no cloud round trip at all. One Pro purchase covers Mac, iPhone and iPad.
Out of Thoth, into your work
Transcripts export as TXT, Markdown, RTF, PDF or JSON, with speaker labels and timestamps you can switch off. SRT and VTT subtitles and a captioned audio video are free for everyone, on every tier. Export the whole library in one pass. Action items from a transcript become real entries in Reminders, only the ones you approve. Share straight to Mail, Messages or AirDrop.
The same scribe, in your pocket
Record and transcribe on device, identify speakers, and run the same AI actions as on the Mac. Import audio from any app. A Live Activity and Dynamic Island keep the recording visible while you use something else. On iPad the library and the transcript sit side by side. Shipped July 2026. Read the announcement →
What I am building now.
One thing at a time. This one is large enough to deserve the whole slot.
Full AI Suite On-Device
Transcription improvement, translation, and AI chat powered entirely by the local model. No internet, no API keys, no data leaving your Mac. On-device summaries already work; the rest of the AI surface needs the local models to hold up on long transcripts, which is the hard part and the reason this gets the whole slot rather than sharing it.
Further out.
Wanted, not yet scheduled. What you vote for is what I look at first.
𓂀 Cast your vote
Which feature matters most to you? One vote per feature. Results are live.
Shared Pad Integration
Push your live transcript to a shared Notion page or collaborative pad in real time, so everyone in the room can follow along on their own screen.
Proton Lumo
Use Proton's privacy-first AI model as a provider in Thoth. Bring your own API key, no data used for training, servers in Switzerland. Waiting on Proton to release their public API.
Remote Ollama
Connect Thoth to an Ollama instance running anywhere: a home server, a NAS, or a remote GPU on Modal, RunPod or similar. Run larger models on dedicated hardware without keeping them on your Mac.
Have a feature request? Reach out at [email protected] or use the feedback button in Settings.
Thoth is built by one person, with no funding, no team, and no investors. Every Pro subscription directly funds the next feature on this list. If you are on the free tier and finding Thoth useful, upgrading to Pro is the most direct way to support its development. Thank you.