Privacy

Last updated 11 October 2026

Product VLM is a voice-first assistant built for blind and low-vision people. This page explains, in plain language, what it keeps and what it shares to answer you.

What we store

• Your account — your email, your name, and your password. The password is stored only as a one-way hash; nobody can read it back.
• Your conversations and preferences — what you asked Mira and what she answered, plus things you asked her to remember, so she can pick up where you left off. Settings such as speech speed are also kept in your browser.
• The latest camera frame of a session, so a follow-up question ("what colour is it?") refers to the same scene.
• Exam Assistant work — question papers you upload, your dictated answers, and their revision history. Examiners and administrators can view submitted exams.
• Face ID, only if you turn it on — a numeric face template (a list of numbers describing your face). Your face picture itself is never stored. Face ID is computed on the Product VLM server, not by a third party, and you can remove it in Settings at any time.

Who helps answer you

To answer, Mira sends only what each step needs to these services:

• Vision — when you ask about what is in front of you, the camera frame and your question go to a vision model through OpenRouter.
• Reasoning — the text of your questions and recent conversation goes to OpenAI.
• Listening — your spoken audio goes to Deepgram to be turned into text (or is transcribed on the server when Deepgram isn't set up).
• Speaking — the text of Mira's answers goes to Google's text-to-speech service to be read aloud.

We don't sell your data and we don't use it for advertising.

Cookies

One cookie keeps you signed in. There are no advertising or tracking cookies.

Your choices

You can change your password and remove Face ID in Settings. To delete your account and its data, ask an administrator — they can do it from the admin console.