How abuzar-ansarii/ultron works
Ultron AI is a voice-first desktop assistant: a user speaks or types a directive, the runtime interprets it, dispatches workstation actions, and presents results through its interface or speech. Its capabilities span system control, proactive context monitoring, gesture input, IoT, web research, and persistent memory. The graph shows the documented workflow and sampled/indexed call relationships; many individual action implementations were not sampled, so their internal dispatch paths are not inferred.
Main entry points: main.py Conversation Runtime · ui.py Desktop Interface · screen_processor.py Screen Processing · action_loader.py Action Loader
Components 29
Assistant Runtime
- Conversation Runtime
main.py - Action Loader
core/action_loader.py - Plugin Loader
core/plugin_loader.py - Configuration
memory/config_manager.py
User Interaction
- Desktop Interface
ui.py - Wake Word
core/wake_word.py - Speech Recognition
core/stt.py - Speech Synthesis
core/tts.py - Audio Devices
core/audio_devices.py - HUD Animation
core/task_animator.py
Workstation Capabilities
- Workstation Tools
actions - Self Evolution
actions/self_evolution.py - Proactive Monitor
actions/chronos_predictive.py - Gesture Control
actions/gesture_control.py - Workspace Modes
actions/workspace_matrix.py - IoT Control
actions/iot_overlord.py - Screen Processing
actions/screen_processor.py - Audio Control
actions/audio_matrix.py
Knowledge and AI
- Web Search
actions/web_search.py - Assistant Memory
memory/memory_manager.py - Cerebro Graph
memory/cerebro_graph.py - LLM Client
core/llm_client.py - Gemini Services
core/gemini.py
Remote Interfaces
- Telegram Control
core/telegram_bot.py - Web Dashboard
dashboard/server.py
Other components
- User
- LLM Providers
- Gemini API
- Telegram API
Connections 33
- User → Desktop Interface: directs
README.md - User → Wake Word: speaks to
README.md - Wake Word → Conversation Runtime: wakes
main.py - Speech Recognition → Conversation Runtime: transcribes for
- Desktop Interface → Conversation Runtime: submits directives
main.py - Conversation Runtime → Action Loader: loads actions
main.py - Conversation Runtime → Plugin Loader: loads plugins
main.py - Action Loader → Workstation Tools: dispatches tools
actions/self_evolution.py - Self Evolution → Action Loader: hot-loads actions
actions/self_evolution.py - Self Evolution → Plugin Loader: hot-loads plugins
actions/self_evolution.py - Conversation Runtime → Assistant Memory: updates memory
main.py - Conversation Runtime → Screen Processing: processes screen
main.py - Conversation Runtime → Web Search: invokes search
main.py - Screen Processing → Gemini Services: analyzes with
actions/screen_processor.py - Screen Processing → Audio Control: uses audio controls
actions/screen_processor.py - Web Search → Gemini Services: uses grounding
actions/web_search.py - LLM Client → LLM Providers: requests completion
core/llm_client.py - Gemini Services → Gemini API: calls
- Conversation Runtime → Configuration: reads settings
main.py - Desktop Interface → Assistant Memory: uses memory
ui.py - Desktop Interface → Speech Synthesis: uses speech
ui.py - Desktop Interface → Audio Devices: selects devices
ui.py - Desktop Interface → Wake Word: uses wake word
ui.py - Desktop Interface → HUD Animation: animates HUD
ui.py - Speech Synthesis → Audio Devices: uses devices
core/tts.py - Conversation Runtime → Telegram Control: initializes control
main.py - Telegram Control → Telegram API: polls and sends
core/telegram_bot.py - Conversation Runtime → Web Dashboard: connects dashboard
main.py - Conversation Runtime → Proactive Monitor: starts monitoring
main.py - Conversation Runtime → Gesture Control: offers gesture tool
- Conversation Runtime → Workspace Modes: dispatches workspace mode
- Conversation Runtime → IoT Control: dispatches room control
- Conversation Runtime → Cerebro Graph: supports graph recall