Skip to content

Devanshupardeshi/DeVA

Repository files navigation

DeVA Logo

πŸ€– DeVA - Your AI Phone Operator

✨ You touch grass. I'll touch your glass. ✨

An open-source AI agent that sees, speaks, and controls your Android phone through voice commands

Features Examples Quick Start Architecture

Kotlin Android Gemini License


🌟 What is DeVA?

DeVA (Device Virtual Assistant) is a revolutionary AI-powered Android assistant that doesn't just respond to your voiceβ€”it actually operates your phone for you. Think of it as having a personal assistant sitting next to you, tapping, swiping, and navigating through apps on your behalf.

πŸ—£οΈ "Hey DeVA, send a good morning text to Mom"
πŸ“± *DeVA opens Messages, finds Mom, types & sends the message*
βœ… "Done! Message sent to Mom"

Why DeVA is Different

Traditional Assistants DeVA
❌ Limited to specific app integrations βœ… Works with ANY app on your phone
❌ Can only answer questions βœ… Actually performs actions
❌ Requires API support from each app βœ… Uses visual UI automation
❌ Closed source, privacy concerns βœ… 100% Open Source

🎯 Example Commands

Task Voice Command
πŸ“§ Send Messages "Send a text to John saying I'll be late"
🎡 Play Music "Open Spotify and play my liked songs"
⏰ Set Alarms "Set an alarm for 7:30 AM tomorrow"
πŸ“Έ Take Photos "Open camera and take a selfie"
🌀️ Check Weather "What's the weather like today?"
πŸ“± Open Apps "Open Instagram and go to my messages"
πŸ” Search "Search for nearby restaurants on Google Maps"
πŸ’¬ LinkedIn "Send welcome message to all new connections"

✨ Features

🧠 Intelligent UI Automation

DeVA sees your screen through accessibility services, understands the context of UI elements, and performs actions like a human wouldβ€”tapping, swiping, and typing.

πŸŽ™οΈ Natural Voice Interaction

High-quality voice recognition and speech synthesis powered by Google's advanced AI. Just speak naturally, and DeVA understands.

πŸ‘οΈ Vision Capability

DeVA can analyze what's on your screen and provide context-aware responses. Ask "What's on my screen?" and get intelligent summaries.

πŸ”Š Wake Word Detection

Say "Hey DeVA" to wake up your assistant anytime, anywhere. No need to open the app first.

πŸ” Privacy First

100% open source. Your voice data stays on your device. No shady data collection.

πŸ“² Works with ANY App

Unlike traditional assistants limited to specific integrations, DeVA can interact with any app on your phone through UI automation.


πŸ—οΈ Architecture

DeVA is built on a sophisticated multi-agent system that separates responsibilities for reliable reasoning:

β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚                     🧠 THE BRAIN (LLM)                      β”‚
β”‚            Gemini-powered reasoning & planning              β”‚
β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€
β”‚                                                             β”‚
β”‚  β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”    β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”    β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”  β”‚
β”‚  β”‚   πŸ‘‚ EARS    β”‚    β”‚   πŸ‘οΈ EYES    β”‚    β”‚   πŸ–οΈ HANDS   β”‚  β”‚
β”‚  β”‚  STT/TTS    │◄──►│ Accessibility │◄──►│   Actions    β”‚  β”‚
β”‚  β”‚   Voice     β”‚    β”‚   Service     β”‚    β”‚  Tap/Swipe   β”‚  β”‚
β”‚  β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜    β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜    β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜  β”‚
β”‚                                                             β”‚
β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€
β”‚                    πŸ“± YOUR ANDROID DEVICE                   β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜

Core Components

  • ConversationalAgentService - Handles voice interactions and conversation flow
  • AgentService - Executes multi-step tasks through UI automation
  • Eyes (Accessibility) - Reads screen content and UI hierarchy
  • SpeechCoordinator - Manages STT/TTS for natural voice interaction
  • GeminiApi - Powers intelligent decision making

⚑ Quick Start

Prerequisites

  • πŸ“± Android device with API Level 26+ (Android 8.0+)
  • πŸ› οΈ Android Studio (latest version recommended)
  • πŸ”‘ Gemini API keys

Installation

  1. Clone the repository

    git clone https://github.com/devanshupardeshi/DeVA.git
    cd DeVA
  2. Configure API Keys

    Create local.properties in the project root:

    # Option 1: Direct Gemini API keys (recommended for testing)
    GEMINI_API_KEYS=your_api_key_1,your_api_key_2
    
    # Option 2: Custom proxy server
    GCLOUD_PROXY_URL=your_backend_url
    GCLOUD_PROXY_URL_KEY=your_password
  3. Build & Run

    • Open in Android Studio
    • Let Gradle sync dependencies
    • Run on your device
  4. Enable Permissions

    • Grant Accessibility Service permission
    • Grant Microphone permission
    • Enable Overlay permission (for floating UI)

🎯 Example Commands

Communication

  • "Text Mom that I'm on my way"
  • "Send an email to my boss"
  • "Call the pizza place"
  • "Reply to the last WhatsApp message"

Entertainment

  • "Play my workout playlist on Spotify"
  • "Open YouTube and play lofi music"
  • "Find a funny video on TikTok"
  • "Open Netflix"

Productivity

  • "Set a timer for 15 minutes"
  • "Create a reminder for tomorrow at 9 AM"
  • "Take a note: buy groceries"
  • "Search for flights to New York"

Navigation

  • "Navigate to the nearest gas station"
  • "How do I get home?"
  • "Find coffee shops near me"
  • "Open Google Maps directions to work"

πŸ› οΈ Tech Stack

Technology Purpose
Kotlin Primary development language
Gemini AI Natural language understanding & decision making
Android Accessibility API Screen reading & UI automation
Firebase Analytics & user management
Google Cloud TTS High-quality voice synthesis
Porcupine Wake word detection

🀝 Contributing

We love contributions! Whether it's:

  • πŸ› Bug reports
  • πŸ’‘ Feature suggestions
  • πŸ“– Documentation improvements
  • πŸ”§ Code contributions

Check out our Contributing Guide to get started.

Quick Contribution Guide

# Fork the repo
# Clone your fork
git clone https://github.com/devanshupardeshi/DeVA.git

# Create a feature branch
git checkout -b feature/amazing-feature

# Make your changes & commit
git commit -m "Add amazing feature"

# Push & create a PR
git push origin feature/amazing-feature

πŸ“œ License

This project is licensed under a Personal Use License.

Use Case Allowed
βœ… Personal & Educational Yes
βœ… Research & Learning Yes
⚠️ Commercial Use Requires separate license

See LICENSE for details.


🌟 Star History

If you find DeVA useful, please consider giving it a ⭐!

Star History Chart


πŸ™ Acknowledgments

  • Built with ❀️ for the open-source community
  • Powered by Google's Gemini AI
  • Inspired by the need for truly accessible technology

Made with πŸ’™ by developers who believe AI should work FOR you

About

AI voice assistant that actually controls your Android phone - Open Source

Topics

Resources

License

Contributing

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors