The hardware

What is a Tactyle screen?

A rugged, all-weather, camera-equipped interactive display — built for shopfronts and public places, not mall-grade indoor panels. On-device AI watches the audience; nothing it sees leaves the device.

Built for the real world

Not a TV. A Tactyle screen.

It starts life as a retail TV panel, but almost everything inside it is rebuilt — a custom control board, a sealed enclosure, and a camera module — so it can survive a shopfront, not just a living room.

The eyes of the screen

It sees the room. It doesn't record it.

On-device AI watches for presence, reads a hand gesture, and estimates anonymous demographics — roughly how many people, roughly what age and gender — and it does all of that computation on the screen itself. No frame of video ever leaves the hardware; only the anonymous result — a count, a gesture, an estimate — goes anywhere at all.

Curious exactly what leaves the screen? See “What our screens see” below.

What plays

More than an ad loop

A Tactyle screen doesn't just play a video on a timer — it runs interactive experiences that respond to the person in front of it, controlled entirely in the air.

Built to withstand the real world

Thrown. Crashed. Intact.

Our durability test, beat by beat: a Tactyle screen gets hurled, hits the ground, and powers back on without a mark.

  1. 01

    Thrown

    A Tactyle screen leaves someone's hands mid-air, tumbling toward the ground.

  2. 02

    Crashed

    Impact — screen-shake, dust, the moment every consumer TV panel would end.

  3. 03

    Intact

    “No cracks. No damage.” The screen powers back on, unmarked, still playing.

Built for outdoors, not just indoors

Trust, made proactive

What our screens see

Screens with cameras earn scrutiny — and they should. Here is exactly what a Tactyle screen sees, computes, and sends, with nothing left implied.

  • On-device, always

    Vision processing runs on the screen itself, not in the cloud. There's no server on the other end watching through the camera.

  • Counts and gestures, never identities

    Presence, dwell time, gestures, and demographic estimates are anonymous events — never linked to a name, a phone number, or a face on file.

  • No facial recognition of individuals

    The vision engine can tell that a face is present and estimate its rough age and gender — it cannot and does not tell you who that person is.

  • Nothing visual leaves the screen

    No photos or video are stored or uploaded, ever. Only anonymous event numbers — a count, a gesture, an estimate — travel to the cloud.

Sample proof-of-play record

The same screen that protects what it sees is built to stay online: remote fleet operations and staged software updates keep it running, day and night.

Try it yourself

Wave at your camera

The same mini screen from earlier on this site — except here, on a desktop with a camera, you can opt in and control it with your actual hand, the way a real Tactyle screen reads a gesture in the air. Everything runs on-device, in your browser; nothing is uploaded.

Likes captured: 0

Tap or click the screen to try the gesture. Keyboard: focus the screen and press Enter or Space. An abstract target appears now and then — swat it before it fades.

Processed on-device. Nothing is uploaded.

Loads ~30 MB vision engine, once, only after you click.

This counter is simulated for demo purposes — it resets on reload and isn't wired to a real screen.

Get started