Overview

One Gesture Engine, Pointed at Ordinary Problems.

Closed beta

Wristcue is a standalone Wear OS app. It watches the accelerometer for seven deliberate motions, and each one can fire any combination of four things: a sound, a spoken phrase, a native watch action such as a media key or launching an app, and an HTTP request to something on your network. Nothing needs a phone in your hand, and the watch does the work itself rather than relaying through a companion.

It came out of CosNFX, which does the same trick for costumes. The gesture engine underneath is shared: the same audio path, HTTP actions, sequences, calibration and web console. What got stripped out was everything costume-shaped, because the interesting discovery was that people kept asking for the engine without the cosplay.

Who It's For

Five Jobs, One Watch

The app ships with seeded profiles rather than an empty screen. Swipe the watch face to switch between them, and the profile's wiring decides what the face becomes: wire a scoreboard and you get a tappable scoreboard, wire several spoken phrases and you get a phrase board.

Communicate

A Voice on Your Wrist

Tap a phrase and the watch says it aloud through on-device text to speech: Yes, No, Help please, Bathroom please. It works in any language the watch has installed, and the phrases are yours to write. This is the profile I care most about getting right.

Coach

Whistle, Horn, Scoreboard

A loud whistle and a stadium air horn on a gesture, spoken calls like Timeout and Water break, a live tappable Home/Away scoreboard that announces the score, and a work/rest interval timer for drills.

Presenter

Slides and a Timer Only You Feel

Flick for next and previous slide, raise to play or pause a video, and run a silent graduated talk timer that buzzes your wrist at milestones. Nobody in the room sees you check the time, because there is nothing to check.

Team Sync

One Watch Cues the Whole Group

A coach's watch can broadcast to the others: a whistle, a horn, a spoken call, a typed phrase or a countdown. An interval timer becomes something a whole class feels on their wrists at the same instant, which is a better way to run a drill than shouting over a gym.

Props & media

Anything With a Local API

Each gesture can also send an HTTP request you define, so it drives WLED strips, Hue, Home Assistant scenes, an ESP32 you built, or an IFTTT applet. Each row has a Test button so you find out it works before you are standing in front of anyone. There's a media remote profile too: play, skip, volume, on the watch or the phone.

The Hard Parts

Gestures Are Personal

A gesture detector that works for the person who wrote it is easy. One that works for a stranger with a different arm, a different watch and a different idea of what "a flick" means is the actual problem.

Calibrate to the Person, Not the Average

Rather than hard-coding which accelerometer axis a motion should show up on, the watch records you performing the gesture and then aims that detector at the channel your motion actually drove. The axes are measured per device instead of assumed, because a watch worn on the other wrist, or a person who flicks from the elbow rather than the wrist, lands the same intent on a different axis.

Configured From a Phone, Served by the Watch

Editing anything complicated on a watch screen is miserable, so the watch runs a small web server and hands the interface to a browser. Open it on any phone on the same Wi‑Fi, enter the six-digit pairing PIN the watch shows, and you get a full control panel: buttons, automations, sequences, sound library, calibration, display and settings. The phone app is the same panel in a wrapper. Nothing leaves the local network.

Seven buttons, and which ones are yours

There are seven detectable gestures: Flick Up, Flick Down, Drop, Push, Grip, Reach and Pull, plus a double-tap that arms and disarms without touching the screen. You choose how many are active and which. Each count from one to seven keeps its own layout, so a three-button setup is a deliberate arrangement rather than the first three of a longer list, while a gesture's sound and actions follow it wherever it appears.

Build & Status

Honest Status

What runs today

The Build

Kotlin and Jetpack Compose on Wear Material 3, running standalone on the watch, with a phone companion also in Kotlin. Built and tested on a Galaxy Watch 7. It ships with a ten-sound starter bank licensed royalty-free, pre-assigned so a fresh install makes noise immediately, and every one of them can be replaced with your own upload.

Not released yet

Where It Is

Wristcue is in closed beta and is not on the Play Store yet, so there is nothing to download from here. The plan is a free download where you buy how many gesture buttons you want active, and owning any one of them unlocks every capability. The site at wristcue.app is live.

On the assistive use

The Communicate profile speaks stored phrases on a tap, which overlaps with what an AAC device does. I want to be careful about that claim: Wristcue is a consumer app on a consumer watch, it has not been evaluated by anyone clinical, and it is not a medical device or a substitute for one. What it is, honestly, is a fast way to put a handful of phrases somewhere you can reach them without a phone.