Automator – Macro & AI Agent is a visual macro builder with on-screen recognition, smart triggers and a built-in AI agent — one app instead of ten utilities. No root. No limits.
Automator turns your phone into an automation platform: build sequences from 190+ actions, act on what's actually on screen via OCR & image matching, fire them on time, location or NFC, and let AI handle the rest.
From a single tap to a full AI-driven workflow across any app — Automator covers the entire surface of your device, no programming required.
Assemble scenarios from steps without code — taps, long-presses, swipes, drag & drop, pinch-zoom, multi-touch, text input and scrolling. Record your actions and play them back. Go beyond clicks with full flow control: conditions, loops, subroutines, try/catch and parallel execution.
Act on what's actually visible. OCR finds and reads text and taps it; image matching locates an element by template; Accessibility reads UI by id, text or coordinates — and waits until it appears.
Set it once, let it run. Launch on a schedule, when you enter a geofenced area, tap an NFC tag, receive a notification, or on system events.
AI helps build and run automation across apps. Bring your own models — cloud (OpenAI, Anthropic Claude, Google Gemini, any OpenAI-compatible endpoint), self-hosted over your LAN (Ollama / LM Studio), or fully offline on-device .gguf models with no internet at all.
Connect Automator to any service by API. Built-in HTTP server, HTTP requests, file downloads, LAN transfer, WebSocket, MQTT client, FTP/SFTP, network device discovery and proxy support.
Variables, lists, tables and a local database (SQLite-style actions). Parse and build JSON/XML, process with regex, encode/decode, hash, encrypt/decrypt, zip archives, manipulate files, and generate or scan QR & barcodes.
Edit PDF and DOCX, process images, drive the camera for photo, video and continuous capture, and stream your device.
Reach the whole phone: SMS (read/send), contacts, calls (dial/answer/end), calendar, notifications, clipboard, Wi-Fi/Bluetooth/flashlight/volume/brightness, screen lock, alarms, location & maps, intents, shell commands, vibration, and battery/app info.
Runs without root. AI keys are stored encrypted. On-device mode works fully offline. And Automator always explains, in plain terms, why it asks for Accessibility.
Connect a cloud provider, point at a model on your own machine, or run a model fully on-device with no network at all. The same agent, image, translation, speech and memory actions work across every mode.
Bring your own key and call the big hosted models directly from a macro. Works with the major providers and anything that speaks the OpenAI-compatible API.
Run open models on your PC or home server and reach them over your LAN. Keep your data on your network and pay nothing per request.
Load a .gguf model straight onto the phone and run it offline.
No key, no server, no connection — the agent keeps working in airplane mode.
No code required. Build visually, test on the spot, then let triggers run it for you in the background.
Automation needs real permissions to work. Here is exactly what Automator uses them for — and what it deliberately does not do.
Android's Accessibility service is what lets an app observe the screen and perform taps and gestures on your behalf. Automator uses it for one purpose: to actually run your macros — reading on-screen content for OCR and image matching, and performing the taps, swipes and inputs your flow describes.
Your automations and data stay with you. There is no root requirement, AI keys are stored encrypted on the device, and on-device mode runs with no network connection at all. Automator does not promise to bypass any platform limit or app protection.
Install from Google Play. 18+. Built for people who want their phone to do the boring parts for them.