How it works

Three states, one gesture

Open you're visible and approachable. Quiet you're visible with a clear "please don't approach". Observer you see only counts — never individuals — and are never visible. Invisible disconnects everything: no socket, no server state.

What you actually see

While walking, open people appear as a soft green marker and quiet people as a dim one — at a glanceable direction and distance category like "about 20–30 m, slightly left". That resolution is deliberate: it's enough to decide whether to say hello, and too coarse to follow anyone. People outside your narrow field of view appear as edge indicators. Text is kept to a minimum; this is a HUD, not a feed.

Mutual visibility

You only ever see individual markers while you are yourself visible (Open or Quiet). There is no way to watch open people while staying unseen — the silent-observer loophole is closed by design. Observers get aggregate counts only, and so does anyone whose location source is too imprecise that moment.

Compass-less alignment

Glasses have no reliable compass, so the display aligns to the real world by walking: after a few straight steps the app compares how far you moved per your GPS track versus your tracked head movement and locks the orientation. Until then it tells you plainly: "alignment low — walk straight for a few seconds."

On the device

The experience is a plain web app at /xr on this site. On Ray-Ban Display glasses you add it once via the Meta AI app in Developer Mode (URL or QR code) — no store, no install. Developers can run the same build on a laptop with the IWER emulator and a simulated walker.