Jarvis

Conversation

Wake word say "jarvis"

Listen passively for "jarvis" or "hey jarvis" and start capturing the rest of your sentence. Energy-gated VAD: audio is only sent to the recogniser when actual speech is detected (not 24/7). Works on iOS Safari, Android Chrome, and desktop. Mic indicator stays on while armed.

Hands-free mode

Auto-listen again after every reply. Tap the mic to stop.

Haptic feedback

Subtle buzz on mic events (iOS may ignore).

Voice

Cloud TTS checking...

Use a high-quality cloud voice (OpenAI / ElevenLabs) instead of the browser's built-in. Configure with OPENAI_API_KEY or ELEVENLABS_API_KEY in .env.

Browser voice

Fallback voice when cloud TTS is off / unavailable.

Speed

Reply speech rate (applies to both browser and cloud).

Memory

Inbox

Quick-capture drops via /api/inbox (text) and /api/inbox/voice (audio → Whisper). Wire from iOS Shortcut or AutoHotkey.

Facts

Long-term facts Jarvis remembers across sessions.

Reminders

Pending pushes Jarvis will fire on schedule.

Watches

Recurring jobs. Jarvis runs the prompt on schedule and only pushes if something needs attention.

Notes

Things Jarvis has jotted down.

Learning

Suggestions

Pending proposals from Jarvis's weekly self-review. Accept to materialise into facts / styles.

Style preferences

How Jarvis should reply, auto-learned from corrections.

Recent activity

Last 20 things Jarvis did (in-memory, resets on restart).

Integrations

Sentry

Recent issues at /api/sentry/issues. Personal token from sentry.io/settings/account/api/auth-tokens (scopes: event:read + org:read).

Slack

Unreads + workspace search. Create an app at api.slack.com/apps, add User Token Scopes (channels/groups/im/mpim:history+read, search:read, users:read), install to workspace, paste the xoxp- token.

System

Push test

Sends a notification to the ntfy topic.