Athena v1.0.0 — Voice-First AI Desktop Assistant
The first stable release of Athena, a voice-first AI assistant that lives on your Windows desktop as a small animated orb bearing the Athena mark.
Ask questions, control local tools, understand what is on your screen, or open a whiteboard — without keeping another chat window in the way. Click the orb or say "Hey Athena" to start speaking.
Install
Download one of the installers attached to this release:
Athena_1.0.0_x64-setup.exe— NSIS installer (recommended)Athena_1.0.0_x64_en-US.msi— MSI installer
Run it, launch Athena, then open Settings to choose an AI provider and enter its API key.
Requires Windows 10 or 11 with the WebView2 Runtime (included with Windows 11 and most current Windows 10 installations).
What's in this release
- Branded orb centerpiece. The orb's plain circular core has been replaced with the branded AthenaMark, while the ambient breathing, ripples, orbiting particles, glow, and speaking volume-pulse animations are preserved.
- Voice interaction. Wake-word listening ("Hey Athena"), speech recognition, spoken responses via local Piper TTS, and continuous conversation mode.
- Screen understanding. Capture the primary screen and ask a vision-capable model to explain it. Screenshots are validated as PNG and capped at 20 MB before analysis.
- Whiteboard. Pen, eraser, pan, undo, redo, zoom, clear, and PNG export — plus optional MediaPipe hand tracking (pinch to draw, two fingers to erase, open hand to move).
- Desktop actions. Open applications, folders, URLs, web searches, and Windows settings; read and write the clipboard; browse files.
- Flexible AI providers. OpenAI-compatible APIs, Anthropic, Google Gemini, Ollama, LM Studio, and custom endpoints.
- Local credentials. Provider API keys are stored in the native OS keyring — never in browser storage.
- Ambient interface. A transparent, always-on-top orb that expands into a glass workspace and closes back into itself.
Verification
This release was built and verified on Windows:
| Check | Result |
|---|---|
npm run build (TypeScript + Vite) |
✅ 63 modules transformed, clean |
cargo check (Rust backend, athena v1.0.0) |
✅ Clean |
npm run tauri build (full release bundle) |
✅ MSI + NSIS installers produced |
| Native screen capture (PowerShell) | ✅ Valid PNG produced |
| PNG magic-byte + 20 MB validation | ✅ Passes validate_screenshot |
| All 37 native commands registered | ✅ Confirmed in invoke_handler |
Interactive features that require a live AI provider key, microphone, camera, or GUI (voice recognition, screen analysis, hand tracking) were verified at the code and build level and compile cleanly, but are not exercised in an unattended build environment.
Privacy
- API keys live in the operating system credential store.
- Screenshots remain local unless you explicitly request screen analysis.
- Athena never silently falls back to another provider.
- Ollama and LM Studio can keep requests entirely on your machine.
Review your provider's privacy policy before sending sensitive screen content to a hosted endpoint.
Athena should feel like an intelligent presence, not another application to manage.