Unseen is the part of your memory that never forgets. An open-source, local-first transcription engine with a memory that compounds: dictate into any app, capture every meeting, ingest your notes — and it quietly distills all of it into a private, structured knowledge base that you and your AI can draw on.
Bring your own AI: Anthropic Claude, OpenAI, Google Gemini, Ollama (fully local), or any OpenAI-compatible endpoint (LM Studio, Groq, OpenRouter, vLLM, …).
Everything stays on your machine. Talk to it all day; forget nothing.
|
Built with ❤️ using Repowise
Unseen was built with Repowise, an open-source code intelligence platform that continuously maps your codebase into structured knowledge for both humans and AI coding agents. Beyond documentation, Repowise continuously monitors code health, highlighting architectural hotspots, files likely to become bug-prone, and refactoring opportunities before they slow development. ⭐ If you're building AI-powered software, check out Repowise: https://github.com/repowise-dev/repowise |
Unseen is one STT + LLM + storage core wearing three hats:
- 🎙 Dictation — tap a hotkey in any app, talk into any text field, get your words inserted at the cursor. Optionally, a fast LLM pass strips fillers, false starts, and stray punctuation first.
- 💬 Meeting copilot — the floating, capture-invisible overlay that transcribes live and surfaces answers, notes, and talking points on demand.
- 🧠 Memory — every dictation, meeting, and note feeds a per-day log that a distillation job turns into an accumulating, structured knowledge base — split into personal and work, combinable on demand. That memory is injected back into your prompts, so the more you use Unseen, the more context your AI has.
dictation ┐
meetings ├─→ daily log (append-only JSONL) ─→ distill ─→ personal / work facts
notes ┘ │
▼
injected as context into your prompts
mic → streaming STT (Deepgram nova-3, diarized)
→ rolling transcript with speaker turns
→ trigger engine (questions, requests, code asks, keywords — per profile)
→ your chosen LLM (streaming, prompt-cached, with your distilled memory)
→ live answer feed in a floating, capture-invisible overlay
Behavior is driven by profiles — plain YAML files that define the prompt, when the copilot speaks up, how it answers, and which memory namespaces it can see:
| Profile | What it does |
|---|---|
| 📝 Meeting Notes | Silent minute-taker: decisions, action items, open questions on demand |
| 💬 Q&A Overlay | Answers any question asked in the room, instantly |
| 📈 Sales Call Assistant | Objection handling grounded in your product docs |
| 🛟 Support Agent | Answers grounded in your documentation, diagnostic steps |
| 🎓 Lecture Companion | Explains jargon and concepts as you listen |
| 🎯 Interview Practice | Mock-interview coach: staged design walkthroughs, real code |
| 💡 Brainstorm Scribe | Clusters ideas, tracks threads, suggests unexplored angles |
Adding your own use case = writing one YAML file. No code.
Press ⌃Space (Control+Space) in any app to start, talk, and press it again to stop. A small HUD shows your words live; on stop the text is pasted at your cursor — clipboard contents restored afterward. If a paste ever fails, the text is left on your clipboard so you can paste it manually.
- Optional AI cleanup (Settings → Memory): off = your words inserted verbatim, instantly; on = a fast model tidies fillers/punctuation first (streamed into the HUD).
- Works in any text field (uses clipboard-paste, not fragile keystroke synthesis).
- macOS needs Accessibility permission once (the first-run wizard walks you through it).
- Per-app exclude list: name apps where dictation should never paste or log.
- The hotkey is remappable in Settings.
Everything transcribed accrues into a private knowledge base — no filing, no manual notes:
- Daily log — dictations, meeting finals, and ingested notes are appended to
memory/log/<date>.jsonl. - Distillation — on demand ("Distill now" in Settings → Memory) or on a schedule, an LLM turns the day's log into deduped, structured facts (people, projects, decisions, preferences, recurring terms). Re-running never duplicates.
- Namespaces — facts are split into personal and work; a profile picks which to inject (
memory.namespaces: [work], or[personal, work]to combine). - Watched sources — point a namespace at markdown files or a folder (work docs, an Obsidian vault) and they're read straight in.
- Apple Notes (macOS) — typed notes are ingested via the official scripting API; handwriting is read from Apple's pre-rendered images and OCR'd locally and offline (swappable engine; see docs/ocr-sidecar.md).
- Background sync (macOS) — Unseen can install its own LaunchAgent to run ingestion + distillation on a schedule, even when the window is closed. No manual setup.
Your data lives on your machine by default. Flip Settings → Memory → Store data in iCloud and Unseen moves memory/, knowledge/, and sessions/ into your iCloud Drive — so a second Mac sees the same logs and memory after cloning. Append-only logs merge cleanly across devices and a per-day lock keeps two Macs from distilling the same day twice. API keys and settings always stay local. The on-disk format is frozen — see docs/data-layout.md.
Download an installer from Releases — dmg (macOS), exe (Windows), AppImage/deb (Linux).
Builds are not yet code-signed: on macOS right-click the app → Open the first time; on Windows click "More info → Run anyway" in SmartScreen.
Or run from source:
git clone https://github.com/repowise-dev/unseen && cd unseen
npm install
npm run devnpm run dist builds the installer for your platform into release/.
First run: open Settings (⚙ in the overlay) → Providers → pick your LLM, paste keys (stored in your OS keychain), Test, done. You need:
- a Deepgram API key for transcription (generous free tier), and
- an LLM: an Anthropic / OpenAI / Gemini key — or no key at all with Ollama running locally.
On macOS, dictation also asks for Accessibility permission so it can paste for you.
| Default | Action |
|---|---|
| ⌃Space | start / stop dictation (talk into any app) |
| ⌘⇧\ | show / hide the overlay |
| ⌘⇧Space | answer the latest thing said ("Ask now") |
| ⌘⇧P | pause / resume listening |
| ⌘⇧] | cycle profile |
| ⌘⇧H | toggle Privacy Mode |
All remappable via hotkeys in settings.json (Ctrl on Windows/Linux).
The overlay is excluded from OS-level screen capture (NSWindow.sharingType = .none on macOS, WDA_EXCLUDEFROMCAPTURE on Windows): your notes don't leak into your screen share. It's a visible toggle (🙈/👁) and you can turn it off. Use it responsibly — see ETHICS.md.
- Transcripts, answers, and your distilled memory stay on your machine (or your own iCloud). No telemetry, ever.
- API keys are stored encrypted in your OS keychain (Electron
safeStorage) and never sync to iCloud. - Handwriting OCR runs fully offline — no cloud, no Claude OCR.
- Pause and a per-app exclude list let you keep specific apps and moments out of memory entirely.
- Recording-consent laws vary by jurisdiction — get consent before transcribing other people. See docs/privacy-and-consent.md.
| Want to add… | Touch | Docs |
|---|---|---|
| A use case | 1 YAML file in your profiles folder | docs/profiles.md |
| An LLM vendor | one LlmProvider implementation |
docs/extending/llm-provider.md |
| An STT vendor | one SttProvider + one parser |
docs/extending/stt-provider.md |
| A trigger detector | one pure function | docs/extending/detector.md |
| An OCR engine | one stdin/path → { text } command |
docs/ocr-sidecar.md |
Architecture overview: docs/architecture.md.
The default pipeline captures your mic. To transcribe the other participants too, route system audio into a virtual device and pick it in Settings → Audio — guide: docs/system-audio.md.
PRs welcome — see CONTRIBUTING.md. Good first contributions: a new profile, a trigger detector, price-table updates, a new provider.