Personal Executive-Function Assistant (AI "Second Brain")

This project aims to create an AI assistant that helps users break down goals, plan tasks, and manage their schedule, acting as a "second brain" for executive function.

This initial version uses BAML for LLM calls and Langgraph for workflow orchestration, running as a Python script.

Running the MVP

1. Setup:
   • Install Python 3.10+
   • Install Poetry: pip install poetry
   • Clone the repository.
   • Navigate to the project directory.
   • Install dependencies: poetry install
2. Environment Variables:
   • Ensure you have a GOOGLE_API_KEY environment variable set, accessible to the script (e.g., in your shell environment or potentially a .env file).
3. Run:
   • Execute the main script: poetry run python main.py

TODO / Future Enhancements

• 1. User Interface (UI):
  • Develop a web interface (e.g., Flask/FastAPI, Streamlit) for user input (goals, tasks, constraints) and schedule visualization.
• 2. Persistence (Memory):
  • Implement storage for user goals, tasks (with status), generated schedules, and preferences.
  • Start with simple file storage (e.g., JSON).
  • Consider a database (SQLite, KuzuDB) for more robust storage.
• 3. Real Calendar Integration:
  • Integrate with Google Calendar/Outlook/Apple Calendar APIs.
  • Read existing calendar events to understand availability.
  • Write generated schedules back to the user's calendar.
  • (Alternative) Generate .ical files for import.
• 4. Dynamic Adaptation & Rescheduling:
  • Allow users to mark task status (completed, delayed, interrupted).
  • Implement automatic rescheduling based on progress/conflicts.
  • Learn and adjust task duration estimates.
• 5. Notifications & Reminders:
  • Add timely notifications (desktop/push) for task starts, switches, and deadlines.
• 6. Improved Task Management:
  • Allow manual task creation, editing, deletion.
  • Add support for specific task deadlines.
  • Enable adding notes/details to tasks.
  • Handle recurring tasks/routines.
  • Explore task dependency visualization (potential KuzuDB use).
• 7. Learning & Personalization:
  • Learn user's focus periods and work patterns.
  • Adjust time estimates based on user performance.
  • Allow user feedback on schedule quality.
• 8. Contextual Awareness (Beyond Tasks):
  • Explore integration with note-taking apps (Obsidian, Notion) or other data sources (emails, docs) for task context (RAG).