Agendly - Executive Function Support Assistant 📅

Agendly is an AI-powered calendar and task management assistant specifically designed for individuals with Executive Function difficulties. It combines calendar management, task scheduling, and voice interaction capabilities to provide comprehensive support for daily planning and organization.

Features

• 🗣️ Voice and text-based interaction
• 📅 Google Calendar integration
• ✅ Smart task scheduling
• ⏲️ Built-in timer for task management
• 🔋 Energy level-aware task planning
• 📚 Context-aware responses using RAG (Retrieval-Augmented Generation)
• 🎯 Automatic task breakdown and scheduling
• 🔄 Real-time calendar updates
• 🗣️ Text-to-speech responses

Prerequisites

• Python 3.12+
• Google Calendar API credentials (Create from Google Cloud Console → Enable Calendar API → Create OAuth 2.0 Client ID → Download credentials.json)
• OpenAI API key
• Ollama (for local LLM support)
• ChromaDB (for document storage)

Installation

1. Clone the repository:

git clone <repository-url>
cd agendly

2. Create and activate a Conda environment:

conda create -n agendly python=3.12
conda activate agendly

3. Install required packages:

pip install -r requirements.txt

4. Set up environment variables: Create a .env file in the root directory with:

MODEL_NAME=<your-openai-model>
OPENAI_API_KEY=<your-openai-api-key>

5. Set up Google Calendar Authentication:

• Place your credentials.json file in the root directory
• On first run, you'll be prompted to authenticate with Google Calendar

Database Setup

1. Create a ChromaDB database for context-aware responses:

jupyter notebook create_database.ipynb

2. Run all cells to initialize the database with your PDF documents

Running the Application

1. Start the Streamlit application:

streamlit run main.py

2. Connect your Google Calendar when prompted

3. The interface provides:

   • Energy level slider (1-10)
   • Voice recording capability
   • Task input and management
   • Calendar event viewing and scheduling
   • Timer functionality

Project Structure

• main.py: Main Streamlit application
• chatbot_with_todo.py: Core chatbot and task management logic
• event_handler.py: Google Calendar integration and event management
• create_database.ipynb: ChromaDB setup for document storage
• requirements.txt: Project dependencies

Features in Detail

Task Scheduling

• Automatically schedules tasks based on:
  • Current energy level
  • Task complexity
  • Available time slots
  • Existing calendar events
• Includes 15-minute breaks between tasks
• Considers user's energy levels for optimal task placement

Voice Interaction

• Records voice input using sounddevice
• Transcribes audio to text
• Provides text-to-speech responses using OpenAI's TTS

Energy-Aware Planning

• Tracks user's energy levels (1-10)
• Adjusts task scheduling based on energy:
  • Low energy (1-3): Shorter, easier tasks
  • Medium energy (4-6): Moderate complexity
  • High energy (7-10): Challenging tasks

Context-Aware Responses

• Uses ChromaDB for document storage
• Implements RAG for informed responses
• Maintains conversation history
• Switches between different LLM models based on needs

Models Used

• OpenAI GPT for main interactions
• Gemma (local) for overwhelm support
• Deepseek for specialized planning
• All-MiniLM-L6-v2 for embeddings

Contributing

[Add contribution guidelines here]

License

[Add license information here]