Autonion is a comprehensive, cross-device AI-powered automation ecosystem designed to seamlessly bridge your Android device, Desktop OS, and Web Browser. Built on an Offline-First, Cloud-Enhanced architecture, Autonion leverages on-device ML, local LLMs, and seamless LAN connectivity for private, rapid execution, while optionally integrating with Cloud APIs for advanced reasoning and expanded capabilities.

The ecosystem consists of four main modules that work in harmony:

The brain and primary interface of the ecosystem. It is an offline-first, cloud-enhanced Android application that empowers users to create powerful automations natively on their phone.

• Key Features: Omni-Chatbot with Langchain and RAG (supporting both local SLMs and Cloud APIs), Gesture Recording & Playback, Flow Automation, Screen Understanding via ML Kit, Semantic Automation, and Hardware Remote capabilities.
• Role: Acts as the central command hub, processing natural language requests via its hybrid AI Engine (Local LLM or Cloud API) and routing them either to local Android execution or to the Desktop Agent.

A Flutter-based desktop application serving as the communication hub for your desktop environment.

• Key Features: Zero-config mDNS discovery, local WebSocket Server bridge, bi-directional clipboard synchronization, Python-based native OS automation, and intelligent task routing.
• Role: Receives agentic action requests from the Android app. It intelligently routes web-related tasks to the browser extension and native desktop tasks to its embedded Python backend.

An AI-powered browser automation extension for Chromium-based desktop browsers.

• Key Features: AI-driven DOM execution, seamless syncing with the Autonion ecosystem, and dedicated content scripts for interacting with LLM interfaces like ChatGPT and Gemini.
• Role: Executes the semantic automation plans inside the browser DOM (e.g., clicking, typing, navigating) as requested by the Autonion-Agent.

A Lemur Browser companion extension for Android devices.

• Key Features: Semantic DOM snapshots extraction, local WebSocket relay, and LLM-driven execution within the mobile browser.
• Role: Acts as a bridge on the Android device itself to allow the Automation Companion app to execute complex web interactions within the mobile browser.

1. User Request: You ask the Automation Companion (Android) via the Omni-Chatbot to perform a task (e.g., "Play my presentation on my PC" or "Log in to my email").
2. AI Processing: The app's hybrid AI Engine (using a local LLM or a configured Cloud API, augmented with RAG/Langchain support) understands the intent and generates an agentic plan.
3. Task Routing:
   • If it's a mobile task, it executes locally (using Accessibility, ScreenML, or the Autonion-Android-Extension).
   • If it's a cross-device task, it communicates via local Wi-Fi to the Autonion-Agent (Desktop).
4. Desktop Execution: The Autonion-Agent receives the command.
   • Native tasks (e.g., opening PowerPoint, media controls) are handled by the embedded Python backend.
   • Web tasks are forwarded to the Autonion-Extension (Desktop Browser), which manipulates the DOM to complete the task.

• Android App: Kotlin, Jetpack Compose, Room, TFLite, ML Kit, Langchain.
• Desktop Agent: Flutter, Dart, Python, mDNS, WebSockets.
• Browser Extensions: JavaScript, Chrome Extension Manifest V3.

Autonion components are licensed under the PolyForm Noncommercial License 1.0.0, permitting personal, research, and educational use.