An AI that sees what you see and answers out loud

Glimpse turns Meta Ray-Ban smart glasses into a real time assistant. You look at something, you ask out loud, and it answers out loud. No wake word, no keyboard, no screen to look at. When the glasses are at home, the same thing runs through your iPhone camera. On the App Store since June 2026.

[ What it is ]
0
Wake words. It is just listening
2
Ways in: the glasses or the phone
es · ja
Localized listings and screenshots, Spanish and Japanese
Role
Founder, Product Designer and iOS Engineer
Timeline
Concept to App Store
Scope
Product design, iOS engineering, Realtime AI, Brand, Marketing site

The challenge

Voice assistants make you talk like a machine

Every assistant on the market wants a wake word, a command shape and a pause in the right place. That is fine on a kitchen speaker and useless on your face. Walking down a street in Shinjuku asking what a sign says, you do not want to say a brand name first and wait for a chime. You want to ask, the way you would ask the person next to you.

What I did

Remove the interface, then make it fast enough that removing it works

There is no wake word and there is no send button. The camera is the input and the voice is the output, and everything in between is a streaming session that has to answer before you have finished forming the next thought. The design work was mostly deletion: no chat log to read, no settings to tune, no screen you need to look at while you are looking at something else. What is left is a listening state and a speaking state.

It also remembers. Say remember this and it keeps the fact, so after a few weeks it knows the coffee you order and the station you use. That is the part I care about most, and it has no screen at all.

← All work