POCO - AI-Powered Personal Assistant

POCO is an AI assistant that transforms WhatsApp into a powerful personal information management platform. Users interact through natural language to store, query, and manage their personal data.

🚀 What It Does

POCO lets you manage your personal information through WhatsApp conversations. Ask it to remember your workouts, track your ideas, manage your tasks, or store any personal data - all through natural language.

Example Conversations:

• "Remember that I worked out for 45 minutes today"
• "What did I do last week?"
• "Add 'Learn Python' to my goals"
• "Show me all my workout sessions from this month"

🏗️ Architecture

WhatsApp → Twilio → FastAPI → LangGraph Agent → MongoDB
                                    ↓
                              AI Tools & Azure Storage

✨ Key Features

• Natural Language Interface: Talk to your data like you would to a friend
• Intelligent Context: Understands time references ("today", "last week")
• Flexible Data Storage: Automatically creates and manages data structures
• Multi-LLM Support: Works with OpenAI, Anthropic, Azure OpenAI, and Groq
• Media Processing: Handles images and documents
• Real-time Responses: Instant WhatsApp integration

🛠️ Technology Stack

Backend & AI:

• FastAPI - Modern async web framework
• LangGraph - AI agent orchestration
• LangChain - LLM integration
• Multiple LLM providers (OpenAI, Anthropic, Azure, Groq)

Database & Storage:

• MongoDB - Document database
• Azure Blob Storage - File storage
• Custom document store with flexible schemas

Infrastructure:

• Docker containerization
• Twilio WhatsApp API
• Azure cloud services

🚀 Quick Start

Prerequisites

• Python 3.12+
• MongoDB
• Twilio account
• OpenAI API key

Setup

1. Clone and Install

git clone https://github.com/oyiakoumis/poco.git
cd poco
pip install -r requirements.txt

2. Configure Environment

# Set up your environment variables
export DATABASE_CONNECTION_STRING="mongodb://localhost:27017"
export OPENAI_API_KEY="your-openai-key"
export TWILIO_ACCOUNT_SID="your-twilio-sid"
export TWILIO_AUTH_TOKEN="your-twilio-token"

3. Run

python src/run_api.py

Docker

docker build -t poco .
docker run -p 8000:8000 --env-file .env poco

📡 API

Main Endpoint

POST /chat/

Processes WhatsApp messages from Twilio webhooks.

Interactive API docs available at http://localhost:8000/docs

🧪 Development

Project Structure

src/
├── agents/           # AI agent and tools
├── api/              # FastAPI application
├── database/         # Data management
└── utils/            # Utilities

Code Quality

black src/ --line-length 160
pytest tests/

🔧 Configuration

Key environment variables:

• DATABASE_CONNECTION_STRING - MongoDB connection
• OPENAI_API_KEY - AI model access
• TWILIO_ACCOUNT_SID - WhatsApp integration
• AZURE_STORAGE_ACCOUNT - File storage

🎯 Technical Highlights

• Async Architecture: High-performance concurrent request handling
• AI Tool Integration: Custom tools for database operations and time resolution
• Distributed Locking: Prevents race conditions in conversations
• Flexible Schema: Dynamic data structure creation and evolution
• Production Ready: Comprehensive error handling and logging
• Cloud Native: Azure integration with secure credential management

📄 License

MIT License - see LICENSE for details.