Who we are

Vixa AI is an assistive-navigation project for blind and low-vision pedestrians. It is currently a working prototype in active development, used by the founding team; it has not yet been released to outside testers. This page covers both this website and the Vixa prototype itself.

Visiting this website

This site is informational. It has no accounts, no sign-up forms, and sets no advertising or tracking cookies of its own. A preference for light or dark mode is stored locally in your browser and never leaves your device. Like almost every website, the server that delivers these pages may keep standard technical logs — IP address, browser type, pages requested — used only to keep the site running and secure.

Third-party services

We may rely on third-party services to operate — for example hosting providers, content-delivery networks, or, in the future, analytics that help us understand how the site is used. Those services may process technical data such as your IP address under their own privacy policies. We choose them with care, but we do not control them, and their terms apply to the data they handle.

We do not sell your data

We do not sell, rent, or trade personal information — not from this website, and not from the Vixa prototype. That is not a temporary position while we are small; it is a design decision. A product whose users trust it with a camera on their face does not get to treat their data as inventory.

Camera data and the Vixa prototype

Vixa works by processing live video of the street in front of you. The phone (or camera glasses, through the phone) streams frames to a guardian server that Vixa runs, where the detection models make one decision, in the moment: does something in your path need your attention? Frames are processed in memory and never stored — not on the server, not on the phone. They are not used to identify people. Frames travel over your local network or a private, encrypted link between your phone and that server. The app also has an explicit, testing-only tunnel mode that sends frames over the public internet to a temporary authenticated endpoint; it is never switched on automatically, and the app tells you what it does before you choose it.

The walk log

The app can keep an optional, text-only log of a walk: the alerts it spoke, connection events, and any “miss” or “false alarm” marks you add yourself. It never contains video or images. It stays on your phone unless you choose to share it with us — and a shared walk log is how testers help us fix detection.

No third-party AI in the guardian path

The collision guardian runs Vixa’s own models on Vixa’s own server; no third-party AI service receives your camera frames in that path. Separately, the prototype contains an experimental conversational assistant that can send a fresh camera frame to a third-party model (Google Gemini) to answer a question you ask. It is off by default, runs only when the server operator has deliberately configured it, and is listed here because a camera product should not have quiet exceptions.

Safety disclaimer

Vixa is an assistive aid and a prototype. It is not a medical device, and it does not replace a white cane, a guide dog, orientation and mobility training, or your own judgment. Silence from Vixa means nothing was flagged — it never means a path has been verified safe. Do not rely on Vixa as your only means of navigating traffic or any other dangerous environment.

Content and trademarks

Product images on this site — such as Ray-Ban Meta glasses and Apple iPhone illustrations — belong to their respective owners and are used here for illustration with source credits. Ray-Ban, Meta, Apple, and iPhone are trademarks of their respective companies, which are not affiliated with and do not endorse Vixa AI.

This website is provided as is

The content of this site is provided for general information, without warranties of any kind. We may change, move, or remove content at any time as the project evolves.

Changes to this page

As Vixa grows from prototype to product, this page will grow with it — and any change will be visible right here. Last updated: September 2026.