ADAM is an AI-powered desktop assistant built by DGEN Technologies Pvt. Ltd.
Powered by the Google Gemini Live API, it features real-time bidirectional voice conversation, emotion-driven face animations, live camera vision & face tracking, async web search, hardware neck servo actuation, and persistent memory.
ADAM/
├── adamV30.py # ★ Latest Entry Launcher (v30)
├── adamV30_test_async_search.py# Core Runtime (v30 async search, vision, neck tracking)
├── adamV29.py # Stable Precursor (v29 search cost optimizations)
├── system_prompt.txt # Unified token-optimized system prompt
├── adam_face.html # Face animation UI (served via Flask/WebSocket)
├── adam_memory.json # Persistent conversation & user memory
├── adam_conversations.json # Conversation transcript logs
├── adam_faces.json # Facial recognition embeddings/data
│
├── emotions/ # HTML/CSS UI assets for emotion state animations
├── faces/ # Reference profile photos for facial recognition
├── tools/ # Diagnostic and verification utilities
│ ├── adam_neck_serial.py # Standalone Arduino/Serial neck test script
│ └── verify_compile.py # Arduino/ESP32 compilation check script
│
├── MP-MC codes/ # Microprocessor / Microcontroller deployment code
│ ├── adam.py # Raspberry Pi / Embedded core daemon
│ ├── laptop_agent.py # Companion agent
│ ├── esp32_cam/ # ESP32 Camera stream firmware
│ └── ADAM_Display/ # Embedded screen drivers
│
├── UNO_code/ # Microcontroller & Arduino firmware
│ ├── picopixel_lvgl_ui/ # LVGL graphical UI for PicoPixel/ESP32
│ ├── esp32_tft_emotions/ # TFT emotion display firmware
│ └── adam_neck_servo/ # Servo neck controller firmware
│
├── mobileAPP/ # Mobile companion application (Turborepo monorepo)
├── adam-web-demo/ # Web demonstration & relay server
├── Hardware_Docs/ # Schematics, blueprints, and setup documentation
├── design/ # Hardware CAD / industrial design assets
└── old_versions/ # Archived historic versions (v9 through v28)
pip install --upgrade google-genai pyaudio python-dotenv websockets flask opencv-python Pillow pyperclip pyserial webrtcvad vosk duckduckgo-search(Optional) Download an offline Vosk speech model from alphacephei.com/vosk/models if using offline wake-word detection.
Create or update your .env file in the project root:
GOOGLE_API_KEY=your_gemini_api_key_herepython adamV30.pyDGEN Technologies Pvt. Ltd. — Kolkata, India
"Innovate. Integrate. Inspire." | Made in India.
- Website: dgentechnologies.com
- Twitter/X: @dgen_tec
- Instagram: @dgen_technologies
- LinkedIn: dgentechnologies