A Unified System for Intelligent Financial Fraud Detection

Modern financial ecosystems require real-time, transparent, and trustworthy fraud detection systems capable of processing massive transaction volumes while meeting regulatory and auditing requirements. This paper presents a unified, modular system integrating four layers: (a) a blockchain-based transaction logging mechanism, (b) an explainable artificial intelligence (XAI) fraud detection engine, (c) a Mistral large language model (LLM) client for natural-language reasoning, and (d) an AI agent for orchestration. A Python-based prototype demonstrates the complete pipeline—from transaction ingestion to fraud scoring, explanation generation, and immutable blockchain storage. The system serves as an educational and experimental platform for research and development, offering flexibility and extensibility for future enhancements.