Chat AI

A standalone NestJS service that handles AI message processing and scheduling for chat applications.

Overview

This service was extracted from the main backend application to provide dedicated AI chat processing capabilities. It includes:

• Scheduler Service: Processes unresponded messages every 10 minutes and cleans up old messages every 12 hours
• AI Message Processor: Determines message topics and generates AI responses using OpenAI
• Database Manager: Manages chat data storage using LMDB
• WhatsApp Service: Handles sending messages (placeholder for actual WhatsApp integration)

Architecture

The service follows a modular NestJS architecture:

src/
├── main.ts                 # Application entry point
├── app.module.ts           # Main application module
├── scheduler/              # Scheduler functionality
│   ├── scheduler.module.ts
│   ├── scheduler.service.ts
│   ├── services/           # Core services
│   ├── interfaces/         # TypeScript interfaces
│   └── constants/          # Configuration constants
└── whatsapp/              # WhatsApp integration
    ├── whatsapp.module.ts
    └── whatsapp.service.ts

Features

Automated Message Processing

• Runs every 10 minutes to process unresponded messages
• Determines message topics using AI
• Generates contextual responses
• Maintains conversation history

Knowledge Management

• Loads topic-specific knowledge dynamically
• Accumulates knowledge across conversation topics
• Supports Portuguese language responses

Database Cleanup

• Automatically removes old messages (older than 12 hours)
• Efficient LMDB storage with compression

Environment Variables

• PORT: Service port (default: 3001)
• LOG_LEVEL: Logging level (default: info)
• OPENAI_API_KEY: OpenAI API key for AI processing

Installation

npm install

Usage

Development

npm run start:dev

Production

npm run build
npm run start:prod

Configuration

The service uses several configuration constants in src/scheduler/constants/topics.ts:

• Topic Keywords: Defines keywords for different conversation topics
• AI Configuration: OpenAI model settings and token limits
• Topic Determination: System prompts for topic classification

Dependencies

Key dependencies include:

• @nestjs/schedule: Cron job scheduling
• @ai-sdk/openai: OpenAI integration
• lmdb: Database storage
• nestjs-pino: Logging

Migration Notes

This service was migrated from the main backend application with the following changes:

• Created standalone WhatsApp service (placeholder)
• Simplified knowledge base retrieval
• Maintained original scheduler functionality
• Added proper TypeScript types for AI SDK

TODO

• Implement actual WhatsApp API integration
• Add comprehensive knowledge base
• Set up proper environment configuration
• Add unit tests
• Add health checks
• Configure proper error handling and retry logic