Building reporting services using Quarkus and AI by Alexis Panousis and Kostas Sidiropoulos

Leverages Quarkus and AI to transform tedious reporting tasks into automated workflows using natural language queries

Uses LangChain4j to integrate with multiple AI models (Ollama, OpenAI) for different purposes:

• Ollama for SQL generation
• OpenAI for security validation
• Ability to compare results between models

Implements guardrails and security features:

• Validates and sanitizes AI-generated SQL queries
• Removes sensitive data like passwords from responses
• Intercepts outgoing payloads for security checks

Key technical components:

• Quarkus for rapid development and hot reload
• Import SQL for sample data generation
• REST API endpoints for natural language queries
• AI models for query translation
• PostgreSQL database backend

Development workflow benefits:

• Reduces boilerplate code writing
• Enables quick iteration through AI assistance
• Maintains developer control while leveraging AI capabilities
• Allows dynamic report generation without constant code changes

Important considerations:

• Need for proper prompt engineering
• Testing AI model outputs
• Balancing convenience vs security
• Understanding model limitations
• Maintaining developer expertise while using AI tools

Built for real business use cases around reporting needs with focus on:

• Sales data
• User information
• Order processing
• Dynamic filtering and aggregation