Handy AI

Handy AI is a voice-driven AI assistant that uses your Android phone for you. Speak a task in plain English -- "Message Sarah on WhatsApp that I'm running late", "Order my usual from DoorDash", "Read me the menu on screen" -- and Handy AI takes a screenshot, asks Claude what to tap next, executes the tap, and repeats until the task is done. It narrates each step out loud, voice-warns before destructive actions, and confirms when it's done. FOR EVERYONE - Hands-free when you're driving or your hands are full - Multi-step tasks in one command instead of a dozen taps - Works in any installed app -- no partner allowlist - Big Stop button kills any task instantly - Your data stays on your device -- no Handy AI server between your phone and Anthropic BUILT FOR ACCESSIBILITY Designed alongside the needs of blind, low-vision, motor-impaired, and cognitive-accessibility users. - Verified TalkBack compatibility -- every button labelled, focus order correct - Text-to-speech reads every assistant reply aloud (auto-on when TalkBack is detected) - Audible action narration -- "Tapping Send. Now in WhatsApp. Message sent." - Voice-warn before destructive actions (Send, Buy, Delete) so you can hit Stop if the agent got it wrong - Easy Mode: one giant mic button, no chat history, no menus - High-contrast text option (pure white on pure black, WCAG AAA) - Haptic feedback for task started, done, error, and confirmation prompts - Set-up-by-computer: get a 6-character code on your phone, enter it on your laptop at gethandyai.app/setup, paste your Anthropic key with your desktop screen reader -- phone receives it - Audit-log voice command: ask "what did you do?" and Handy AI reads back the last few actions - Bluetooth headset + hearing aid compatible Handy AI combines open-ended automation with verified accessibility-tool status -- declared as an accessibility service in the Android system settings, not just marketed as one. Other apps (Be My AI, Seeing AI, Lookout, Envision) describe your screen -- Handy AI executes the task on it. HOW IT WORKS 1. Tap the mic, speak a task. Or type it. 2. Handy AI screenshots your screen. 3. Sends the screenshot to Claude with your task. 4. Claude returns the next action (tap, type, swipe, etc.). 5. Handy AI executes the action via the Accessibility API. 6. Repeat from step 2 until done. REQUIREMENTS - Android 11 or higher - Anthropic API key (free $5 credit at console.anthropic.com) - Accessibility permission (granted once, revoke anytime) PRIVACY Your Anthropic key lives encrypted on your phone only. Screenshots go directly from your phone to Anthropic's Claude API -- never to a Handy AI server. We store nothing, train on nothing, proxy nothing. KEYWORDS blind, low vision, screen reader, TalkBack, voice control, hands free, motor accessibility, ALS, Parkinson's, vision impaired, voice command, voice assistant, accessibility AI, AI agent, Claude, Anthropic, phone automation, Easy Mode, multi-step tasks, AI accessibility SOURCE & SUPPORT Open source: github.com/fainir/handy-ai Email: [email protected] Privacy: gethandyai.app/privacy.html