Hi everyone,
I've been building an iPhone app for a while, mostly because I wanted something like it myself, and I think it has finally reached the point where I want other people to start poking holes in it.
The simplest way I can describe it is an assistant that combines a conversational camera with location, navigation and public transport tools.
I know there are already a lot of AI visual-assistance apps around. What kept bothering me wasn't that they couldn't recognise things. A lot of them can. It was the interaction.
You ask what's in front of you, get an answer, move the camera, ask again, and quite often you're effectively starting over.
Even with newer apps that can keep the camera running, the visual assistant, navigation, nearby places and public transport can still feel like separate tools.
What I've been trying to build is something that stays with the situation.
If I'm looking through a shelf, I want to keep exploring it.
If I'm trying to read tiny writing on a perfume bottle, I want to keep working with the same object.
I might ask:
"What's on the top shelf?"
"Which bottle am I holding now?"
"What shops are around here?"
"What does the sunset look like?"
"What's happening on the TV?"
Or I might tell it to stay quiet and only let me know when something I'm waiting for happens, such as my bus arriving.
The important part for me isn't simply that the camera is live. It's keeping the camera, conversation and what I'm actually trying to do connected.
I've been applying the same idea to navigation and public transport.
I'm not trying to replace a cane, guide dog or mobility skills with AI, and I'm not particularly interested in building another GPS voice that only says "turn left in 100 metres."
I'm more interested in questions like:
Which side of the street am I on?
What's around me relative to the direction I'm facing?
Where's the next corner?
Is the place I'm looking for coming up on my left or right?
Which nearby stop is actually mine?
What is leaving from it?
And once I'm on the bus, why should following the journey become a completely separate workflow?
Most of my navigation and transit testing is currently around Istanbul. Some parts work well enough that I use them regularly; other parts are still experimental.
I've also been using Meta Ray-Bans as another camera source. Having the camera roughly where you're already looking makes continuous assistance much more natural than holding the phone up for everything.
There is also an offline assistant running directly on the iPhone. It isn't as capable as the online side, but basic visual questions, reading and finding objects can work without sending the camera anywhere.
The app is currently iPhone-only, supports VoiceOver throughout, and works in English and Turkish.
I'm deliberately not listing every feature here.
I'm also definitely not claiming this is some revolutionary system that "solves blindness." It doesn't.
What I want is something much less dramatic: a reliable source of information I can reach for, ask follow-up questions about, and then get on with whatever I was doing.
If it can save me from switching between several apps, repeatedly taking pictures, or asking the same question five different ways, that's already useful.
I've been using it myself and a few friends have been testing it, but we're still a tiny group. I've inevitably designed parts of it around how I personally use a camera, navigate and use public transport.
So I'd like to hear where other people's needs differ.
I've made a short survey here: Survey link
It should take around 3 to 5 minutes.
I'm especially interested in where current visual-assistance apps make you repeat yourself, where navigation or public transport apps leave gaps, when proactive speech would be useful or annoying, and what very specific visual tasks current apps still handle badly.
"I wouldn't use this because..." is useful feedback too.
I'm also going to start inviting a small number of people to a TestFlight build rather than posting a public link immediately.
There's an optional place in the survey to leave an email address if you're interested in testing. It'll only be used for TestFlight invitations.
For now I'm mainly looking for real use cases, complaints, strange ideas and places where the whole concept falls apart.