Wristotle Companion
- Categories are now grouped under section headers
(Watch, Voice & AI, Features, Data & system, About),
with a few related screens folded together for fewer
top-level rows.
- A search box filters Settings by name or keyword.
- Busy sub-menus carry thin theme labels to scan faster.
- Prefer the old flat list? A toggle switches back to
the classic layout (grouped stays the default).
Everything runs on-device by default. Optional cloud Whisper / self-hosted speech-to-text endpoint and optional Anthropic or OpenAI-compatible LLM if you want them. Baseline doesn't phone home; no account, no tracking, AGPLv3-licensed.
Works with microPebble and Core Devices (rePebble).
Voice commands (one-liners):
* Calls — "call John"
* Messages — "text Sarah I'll be late" / "whatsapp Mom dinner at 7"
* Reminders — "remind me to take meds in 20 minutes" (incl. persistent "nag me" reminders)
* Meetings — "schedule a meeting with Sam tomorrow at 3" / "what's on my calendar"
* Alarms — "set an alarm for 6:30 am" (phone / watch / both)
* Timers — "set a timer for 10 minutes"
* Music — "play Spotify" / "pause" / "next"
* Apps — "open Maps" / "launch Audible"
* Notes — "make a note buy milk" (text + audio)
* Tasks — "add task pick up dry cleaning"
* Weather — "weather in Tokyo" / "what's the weather"
* Sports — "when do the Warriors play next" / "did Arsenal win"
* World clock — "what time is it in Berlin"
* Calculator — "what's 15% of 80"
* Morning brief — "what's my morning brief"
* Ask Agent — "ask agent what's on my calendar tomorrow"
* Find phone — "find my phone"
A first-launch welcome wizard walks new users through the optional setup that unlocks the full app — Whisper speech model download, on-device intent classifier model, runtime permissions, installed-app indexing.
Other things worth knowing:
* Watch cards — sports, weather, and meeting replies render as full-screen cards on the watch (colour band shows the action, icon shows the subject); pick which kinds show in settings
* Speak replies on the watch — optional on-device or cloud text-to-speech reads answers aloud on supported watches
* Confirm-before-send for destructive actions (toggle in settings or say "confirm" in your query) previews destination + body before any call / message goes out
* Conversation history with optional audio capture
* Contact aliases — "call dad" stores the mapping so it survives phonebook re-orders
* App aliases — "play tunes" -> Spotify
* Custom Ask-Agent triggers — "Jarvis, what's the weather"
* STT providers — on-device Whisper, OpenAI-compatible HTTP endpoint (Groq, Cloudflare, Speaches, whisper.cpp-server, ...), or composite failover
* Backup & Restore — per-entity JSON ZIP with optional AES-256
* MCP client — Streaming-HTTP-by-default with SSE fallback; the Ask-Agent loop aggregates tools across enabled servers
You'll need the Wristotle watch app on your Pebble too — install it from the Rebble appstore (microPebble) or the Core Devices appstore.
Full feature list, install guide, and troubleshooting at https://wristotle.codeberg.page/
- Author: Lazy Devs
- License: GNU Affero General Public License v3.0 only
- Website
- Issue Tracker
- Source Code
- Changelog
- Build Metadata
- Reproducibility Status
Versions
Although APK downloads are available below to give you the choice, you should be aware that by installing that way you will not receive update notifications and it's a less secure way to download. We recommend that you install the F-Droid client and use that.
Download F-Droid-
arm64-v8aThis version requires Android 7.0 or newer.
It is built and signed by the original developer, and guaranteed to correspond to this source tarball.
Download APK 32 MiB PGP Signature | Build Log
-
arm64-v8aThis version requires Android 7.0 or newer.
It is built and signed by the original developer, and guaranteed to correspond to this source tarball.
Download APK 32 MiB PGP Signature | Build Log
-
arm64-v8aThis version requires Android 7.0 or newer.
It is built and signed by the original developer, and guaranteed to correspond to this source tarball.
Download APK 32 MiB PGP Signature | Build Log
- Settings now opens to a short core list, with a
"Show advanced settings" toggle for the optional extras
(integrations, diagnostics).
- The first-run guide downloads the models it needs with
one tap — no digging into Settings — and ends with a
quick "try it on your watch" check.
- "Send unrecognised speech to Ask Agent" routes anything
the app doesn't recognise to your agent, while keeping
phone commands (call, text, timers) working.
- "Ask Agent only" sends every spoken command straight to
your agent, with no wake word needed.
Both are off by default and need a provider configured.