๐Ÿง 

Autonomous Planning & Re-evaluation

Give OpenDroid a complex, multi-step goal like "Check if it's going to rain, text my wife I'll be late, and set an alarm for 6 PM" โ€” and it handles everything. The agent decomposes commands into logical sub-tasks, executes each in sequence, verifies the outcome, and dynamically adjusts the remaining steps if something fails or conditions change.

PlanManager ReEvaluationEngine IntentClassifier Compound Intent Guard
๐Ÿ“ฑ

Full Device & System Control

OpenDroid provides deep integration with Android system APIs, enabling it to control brightness, toggle Wi-Fi and Bluetooth, manage the flashlight, lock the screen, set alarms and timers, create calendar events, translate languages, convert currencies, and much more โ€” all without opening a single app.

SystemActions WRITE_SETTINGS AlarmManager CalendarProvider
๐Ÿค–

Accessibility Automation

When API controls aren't available, OpenDroid leverages Android's Accessibility Service to interact with any app directly. It can click buttons, scroll lists, read screen content, type text, and navigate complex UIs โ€” enabling tasks like sending WhatsApp messages, finding locations in Maps, or posting on social media.

AccessibilityService UI Automation WhatsApp Automator Screen Reader
๐Ÿ“ธ

Vision Engine (Screenshot Analysis)

Fully integrated multimodal vision. OpenDroid captures the screen buffer via the Accessibility API (Android 11+) and sends it to the LLM for visual understanding. On older devices or when permissions are limited, it automatically falls back to an accessibility text-scraping mode โ€” ensuring seamless operation across all Android versions.

Multimodal LLM Screen Capture Auto Fallback Android 11+
๐Ÿ—„๏ธ

Multi-Tier Persistent Memory

OpenDroid doesn't forget. Its four-tier memory system includes: Working Memory for current task context, Episodic Memory for past action logs, Semantic Memory for long-term personal facts extracted via LLM-powered fact mining, and Procedural Memory for custom user-defined macro workflows.

Working Memory Episodic Memory Semantic Memory Procedural Memory
๐ŸŽ™๏ธ

Wake Word & Voice Interface

Go completely hands-free with offline wake word detection, Android's native speech recognition engine, and high-fidelity text-to-speech output. OpenDroid also supports premium ElevenLabs voices as a TTS fallback for a more natural conversational experience.

Wake Word Speech-to-Text Text-to-Speech ElevenLabs
๐Ÿ“ž

Robust Calls & SMS

Zero-refusal communication with intelligent fallback chains. OpenDroid resolves contacts by name, makes calls via TelephonyManager, and sends SMS programmatically. If permissions are missing, it gracefully falls back to system dialer and SMS composer intents โ€” ensuring the task always gets done.

ContactResolver TelephonyManager SmsManager Intent Fallback
๐Ÿ”—

10+ LLM Provider Support

Bring your own AI brain. OpenDroid supports Anthropic Claude, OpenAI GPT, Google Gemini, Groq, Mistral, OpenRouter, Together AI, Cohere, DeepSeek โ€” and fully offline operation via Ollama with a custom server URL. Switch providers instantly from the settings screen.

Anthropic OpenAI Gemini Ollama (Offline) +6 More
๐ŸŽจ

Premium Glassmorphic Design

Built with Jetpack Compose featuring a futuristic deep navy and neon green design system. Enjoy custom pulsing audio orb indicators, live latency benchmarks, animated plan visualizations, and a polished chat interface that makes AI interaction feel premium.

Jetpack Compose Dark Theme Glassmorphism Animations

Experience Every Feature

Download OpenDroid and explore its full capabilities on your own Android device.