A hierarchical agent-based approach to email management that processes emails in layers instead of feeding an entire inbox to a single LLM at once.

The system consists of specialized components arranged in a hierarchical structure:

1. Cognitive Email Ecosystem: Core system that orchestrates multiple agent layers
2. Ingestion Agent: Normalizes email data for processing
3. Observer Agent: Categorizes emails and detects user traits
4. Context Agent: Analyzes external factors affecting email interpretation
5. Social Graph Agent: Maps and understands relationship networks
6. Intent Decoder: Determines the purpose and meaning behind emails
7. Execution Specialists: Handles specific email types with specialized logic

.
├── cognitive_email_ecosystem.py  # Core hierarchical agent system
├── src/
│   ├── ingestionAgent.py         # Email data loading and normalization
│   ├── observerAgent.py          # Email categorization and user trait detection
│   ├── cognitive_email_adapter.py # Connects agents to the cognitive system
├── tests/                        # Unit tests
├── email_interface.py            # Flask web interface
├── observe_demo.py               # Demonstration script
├── data/
│   ├── syntheticEmails.json      # Sample email threads
│   ├── observerSessionData.json  # Thread data for the Observer Agent
│   ├── observerLongTermData.json # Long-term user trait storage

• Multi-layered email processing pipeline
• Email categorization into meaningful buckets (Work, Social, Bills, etc.)
• User trait detection based on email patterns
• Relationship mapping through social graph analysis
• Web interface for visualization and interaction

1. Clone the repository
2. Install dependencies
3. Run the demonstration script:

   python observe_demo.py
   

4. Access the web interface:

   python email_interface.py
   

The hierarchical agent-based approach provides:

• More targeted analysis through specialized components
• Better understanding of context and relationships
• More efficient processing by breaking down the task into manageable layers
• Improved user experience through adaptive categorization