Lumi — AI vision assistant

Lumi is an AI assistant that reads what is on your screen and acts on your behalf in other apps — designed for users for whom standard touch-and-type interaction is difficult, slow, or impossible.

WHO LUMI IS FOR

Lumi is designed first for people who have a hard time using a standard touch-screen phone. That includes:

• People with low vision or blindness, who need their phone to describe what's on screen and act on their spoken instructions rather than relying on visual scanning and precise touch.
• People with motor impairments — tremor, limited dexterity, one-handed use, hands occupied with care work — who need synthetic gestures dispatched to apps rather than manual taps and swipes.
• People with cognitive-load constraints — executive-function challenges, reading fatigue, language barriers — who need long or visually dense screens summarised, and multi-step flows walked through one prompt at a time.

If standard touch-and-type works for you, Lumi will still feel useful — but it is built first for the people above.

WHAT LUMI DOES

• Reads any screen aloud. Open the Lumi mini-chat overlay, ask "what is on this page?", and Lumi describes the foreground app in plain language — long-form content, settings pages, dense lists, all of it. Useful when the screen is too small, the contrast is wrong, or the font is unreadable.

• Summarises long content. "Summarise this email", "give me the three things on this page" — Lumi pulls the essentials out of whatever is on screen so you don't have to scan it all.

• Acts on your spoken or typed instruction. "Tap Reply", "scroll down", "type 'on my way' and send", "go back", "go home" — Lumi dispatches the gesture or keystroke to the app you're in. No reaching, no holding, no fine-motor precision required.

• Opens apps for you. "Open the calendar", "launch Maps" — Lumi launches apps without you needing to find the icon.

• Reduces step-by-step friction. Lumi walks you through a multi-screen task — booking, replying, navigating a settings flow — one spoken prompt at a time. You don't have to remember each step.

HOW LUMI USES ACCESSIBILITY

Lumi uses Android's accessibility service so it can read the foreground app's content and act in it on your behalf. The accessibility service is the only Android API that gives a non-system app both halves of that experience.

When you send a message in the Lumi mini-chat overlay, Lumi captures the foreground app's structure, a screenshot, and the package name, and sends them with your message so the AI can ground its reply in what you're looking at and act if you ask it to. Because Lumi reads the screen, the captured content can include other user-generated content shown in the app you're using — messages, emails, documents, and posts written by you or by other people. This is sent to Lumi's AI only to answer the request you just made.

We never capture proactively, never in the background, never while the screen is off. We do not capture your keystrokes. You can revoke the permission at any time in Settings → Accessibility → Installed services → Lumi, or from Lumi's in-app Accessibility settings row.

DATA AND PRIVACY

What Lumi captures via the accessibility service is described in plain language in the Lumi Privacy Policy: https://lumi.nexoragroup.dev/privacy-policy/#accessibility — including what is captured, when, where it goes, and how long we keep it. Lumi does not sell your data, does not use it for advertising, and does not retain it for AI training.

NEED HELP?

Email lumi-support@nexoragroup.dev. We answer within one business day.