Augen: Instant Answers About What You See
An accessibility-first visual AI concept: point the phone, ask, get a short answer. No chat, no signup. Speed is the whole point.
Augen (German for “eyes”) is a project concept I keep coming back to: a visual question tool for one specific moment. A blind or low-vision person points their phone at something and needs to know, right now, what it is or what it says.
Every existing assistant gets in the way of that moment. You open an app, wait for it to load, maybe create an account, start a chat, describe what you want, wait for a verbose answer. Any one of those steps is fine. All of them together, when you just need to know what a sign says, are too much.
The idea
Augen is one screen. Take a photo or point the camera live, ask by voice or text (“what does this say”, “is this the right medicine”, “what color is this”), and get a short answer immediately.
No chat history, no conversation context to manage, no account. Using it a hundred times a day should feel like glancing at something.
For longer interactions there would be AugenChat, a companion for conversations that need context: reviewing a document, comparing labels, learning a room layout. Augen itself stays deliberately dumb and fast.
Why now
This isn’t hypothetical demand. Blind or low-vision people already strap camera glasses to their heads and ask questions out loud. The Ray-Ban Meta glasses turned out to be an accidental accessibility device, and the community noticed before Meta did.
What doesn’t exist yet is an ecosystem that puts accessibility first instead of bolting it on:
- Open, so users aren’t locked into one vendor’s glasses and one vendor’s models.
- Local where possible, because a photo of your medication shouldn’t need an upload.
- Accessibility by default. The settings exist for everyone, the interface assumes them, and temporary situations - a broken arm, glare on a sunny street - stop being edge cases.
Status
This is a concept I’m shaping into a build. The core loop is small enough to prototype quickly: capture, one vision model, one question, one answer. Most of the work is cutting out every unnecessary tap and millisecond.
More as it takes form. If you work in accessibility or use these tools daily, I’d like to hear what breaks in your current workflow.
Drafted with AI, edited by hand. I publish it when it sounds like me.