How tarigopullasathwik/anju_ai works
Anju AI is a desktop-oriented conversational assistant with a Flask/WebSocket dashboard, voice and file capabilities, persistent memory, and actionable tools. The main workflow is user input through the dashboard to the assistant brain, which can answer or plan multi-step work and dispatch actions. Memory supports personalized context; tools cover search, file/document, image, camera, code, phone, and system tasks. The excerpts establish the main entry point, workflow routing, plugin fallback, memory persistence, and model client; links among unsampled subsystems are intentionally limited.
Main entry points: workflow_engine.py Workflow Engine · brain.py Conversation Brain · plugin_manager.py Plugin Manager · main.py Assistant Runtime
Components 30
User Interaction
- Web Dashboard
utils/ui_dashboard.py - Assistant Runtime
main.py - Voice Input
voice/listen.py - Wake Word
voice/wake_word.py - Speech Output
voice/tts.py - Neural TTS
voice/free_tts.py
Assistant Intelligence
- Conversation Brain
brain/brain.py - Task Planner
brain/task_planner.py - Workflow Engine
brain/workflow_engine.py - Sentiment Analysis
emotion/sentiment_analysis.py - Model and API Routing
utils/api_handler.py
Memory
- Memory Context
brain/memory_manager.py - Memory Operations
memory/memory.py - SQLite Memory Store
memory/database.py
Tools and Actions
- Plugin Manager
tools/plugin_manager.py - Web Search
tools/web_search.py - File Operations
tools/file_ops.py - Document Tools
tools/file_processor.py - PDF Generation
tools/pdf_generator.py - Image Generation
tools/image_generator.py - Camera
tools/camera.py - Code Execution
tools/code_execution.py - Phone Control
tools/phone_control.py - Desktop Automation
automation/app_control.py - System Tasks
automation/system_tasks.py
External Services
- TokenRouter Client
utils/tokenrouter_client.py - LLM Provider
Other components
- User
- Web Search Service
- Edge TTS Service
Connections 34
- User → Web Dashboard: sends input
- Web Dashboard → Assistant Runtime: delivers command
- Assistant Runtime → Conversation Brain: processes query
- Assistant Runtime → Web Dashboard: updates status
- Assistant Runtime → Conversation Brain: handles uploads
- Voice Input → Conversation Brain: provides speech
- Wake Word → Voice Input: activates listening
- Conversation Brain → Speech Output: produces speech
- Conversation Brain → Neural TTS: requests synthesis
- Neural TTS → Edge TTS Service: requests audio
- Conversation Brain → Task Planner: plans task
- Workflow Engine → Task Planner: plans command
- Conversation Brain → Model and API Routing: requests model
- Model and API Routing → TokenRouter Client: sends completion
- TokenRouter Client → LLM Provider: calls API
- Conversation Brain → Memory Context: builds context
- Memory Context → SQLite Memory Store: reads and updates
- Memory Operations → SQLite Memory Store: reads and writes
- Model and API Routing → Memory Operations: searches memory
- Workflow Engine → Web Search: dispatches action
- Web Search → Web Search Service: searches web
- Workflow Engine → PDF Generation: dispatches action
- Workflow Engine → Image Generation: dispatches action
- Workflow Engine → Camera: dispatches action
- Workflow Engine → Code Execution: dispatches action
- Workflow Engine → File Operations: dispatches action
- Workflow Engine → Phone Control: dispatches action
- Workflow Engine → Desktop Automation: dispatches action
- Plugin Manager → Workflow Engine: falls back to
- Plugin Manager → Web Search: registers builtin
- Plugin Manager → Image Generation: registers builtin
- Plugin Manager → PDF Generation: registers builtin
- Sentiment Analysis → Conversation Brain: informs response
- Workflow Engine → System Tasks: dispatches action