An intelligent chat system for grant identification and evaluation, designed to help sales representatives and customers find and assess funding opportunities.

🎯 Main Objective

Develop an intelligent chat system that allows sales representatives and clients to:

• Identify available grants
• Evaluate eligibility for specific projects
• Determine the suitability of grants
• Maximize possible financing returns
• Interact naturally with queries about grants

🏗️ Architecture

The system consists of the following main components:

Frontend

• Interactive chat interface
• Developed in React with Material Tailwind
• Responsive design
• Authentication system
• Markdown rendering for rich responses

Backend

• FastAPI containerized services
• Integration with Fandit grants API
• LangGraph state machine for conversation flow
• ETL system for data management
• Intermediate database for efficient filtering

AWS Services

• Bedrock: AI engine for natural language processing
• DynamoDB: High-speed database for chat history
• Aurora: MySQL database for grant information
• EC2: Instances for service deployment

⚡ Key Features

1. Intelligent Chat Interface

   • Natural language interaction
   • Contextualized responses
   • Automatic eligibility assessment
   • Conversation history

2. Data Management

   • Integration with Fandit API
   • Intermediate database for efficient filtering
   • Automatic information updates via ETL
   • TTL-based expiration for conversation data

3. Grant Analysis

   • Filtering by region, company type, and budget
   • Detailed grant information
   • Eligibility criteria matching
   • Grant comparison

4. Session Management

   • Concurrent user sessions
   • Conversation persistence
   • Automatic cleanup of inactive sessions
   • Chat history retrieval

🚀 Technical Requirements

Functional

• User authentication
• Natural language interaction
• Advanced filtering system
• Intermediate database
• Query processing
• Conversation management

Non-Functional

• High availability
• Optimized performance
• Responsive design
• Data security
• Scalability

💻 Technologies

• Frontend: React, Material Tailwind, Markdoc
• Backend: Python, FastAPI, LangGraph
• AI: AWS Bedrock
• Databases: Aurora MySQL, DynamoDB
• ETL: Custom Python pipeline
• Authentication: Context-based with JWT
• Containers: Docker
• Deployment: Docker Compose

🛠️ Installation and Development

Prerequisites

• AWS Account
• Docker and Docker Compose
• Node.js and npm
• Python 3.9+
• Git

Environment Configuration

1. Clone the repository

   git clone https://github.com/AI-School-F5-P3/ChatbotGrants.git
   cd ChatbotGrants

2. Configure environment variables

   # In backend directory
   cp .env.example .env
   # Edit .env with your credentials
   
   # In etl_fandit directory
   cp .env.example .env
   # Edit .env with your Fandit API credentials

3. Start the services

   docker-compose up -d

Project Structure

ChatbotGrants/
├── README.md                   # This documentation
├── docker-compose.yml          # Main Docker composition file
├── requirements.txt            # Root level dependencies
│
├── backend/                    # Backend Python code
│   ├── aws_connect.py          # AWS services connection module
│   ├── Dockerfile              # Backend container definition
│   ├── dynamodb.py             # DynamoDB integration for chat history
│   ├── grants_bot.py           # Core chatbot logic using LangGraph
│   ├── main.py                 # FastAPI application with endpoints
│   ├── requirements.txt        # Backend dependencies
│   └── tools_aurora.py         # Database operations for grants
│
├── frontend/                   # React application
│   ├── Dockerfile              # Frontend container definition
│   ├── package.json            # Frontend dependencies
│   ├── public/                 # Static assets
│   │   └── img/                # Image assets
│   └── src/                    # Source code
│       ├── components/         # UI components
│       ├── context/            # React contexts
│       ├── pages/              # Application pages
│       ├── services/           # API services
│       └── assets/             # Styles and images
│
└── etl_fandit/                 # ETL system for Fandit API
    ├── clase_apifandit.py      # Fandit API client
    ├── db_setup.py             # Database initialization
    ├── etl_fandit.py           # Main ETL process implementation
    ├── Dockerfile              # ETL container definition
    ├── requirements.txt        # ETL dependencies
    ├── README.md               # ETL documentation
    └── tests/                  # ETL test directory

📝 API and Resources

Fandit API

• Endpoint: https://ayming.api.fandit.es/api/v2
• Update frequency: Daily
• Grant filters supported:
  • is_open
  • start_date
  • end_date
  • provinces
  • communities
  • action_items
  • activities
  • region_types

Backend API Endpoints

• /start_session: Initialize a chat session
• /chat: Process messages
• /end_session/{user_id}: End a user session
• /save_chat: Save conversation to DynamoDB
• /get_chat_messages: Get messages for a conversation
• /get_user_conversations/{user_id}: List user conversations

🔄 Workflow

1. User Input

   • Natural language query
   • Basic information collection
   • Grant selection

2. Processing

   • Query analysis
   • Database search
   • Eligibility evaluation
   • LangGraph state transitions

3. Response

   • Relevant grants
   • Suitability analysis
   • Recommendations
   • Rich text formatting

📊 Monitoring and Maintenance

• System logs
  • Backend logs
  • ETL process logs
  • DynamoDB TTL enforcement
• API quota control
• Database backups
• Inactive session cleanup

👥 Team and Support

• Frontend Development
• Backend Development
• DevOps
• Business Team

📫 Contact and Help

For support and inquiries:

• Create an issue in the repository
• Contact the development team
• Consult internal documentation

📜 License

This is a public educational project.