Agentic AI in Financial Systems: Architectures, Compliance Automation, Fraud Detection, and Smart Contract Orchestration

This literature review synthesises current research on the integration of agentic AI within financial systems, with a focus on AI architectures, compliance automation, fraud detection, and smart contract orchestration. The review aims to assess the reported effectiveness of agentic AI technologies in improving operational efficiency, supporting regulatory compliance, detecting financial fraud, and facilitating smart contract execution. Existing studies indicate that financial institutions are progressively exploring agentic AI architectures to enhance both front-end and back-end operations, with some evidence suggesting potential benefits in automating compliance processes and reducing regulatory violations. Additionally, agentic AI shows promise in identifying patterns related to financial fraud, though findings on real-world effectiveness vary. Research on AI-driven smart contract orchestration points to improvements in transactional speed and accuracy, albeit with limited empirical data. Cost-effectiveness comparisons between traditional and AI-agentic methods remain preliminary, with few studies providing comprehensive analyses over extended periods. This review highlights the emerging role of agentic AI in financial systems while identifying gaps and limitations in the current literature, highlighting the need for further empirical research to validate and expand upon these initial findings.