Financial institutions are rapidly advancing in their use of Generative AI (GenAI), driven by its potential to enhance investigations, automation, and service delivery, yet face challenges in scaling it responsibly due to legacy systems, complex AI stacks, and regulatory requirements. The adoption of GenAI signifies a shift from data analysis tools to systems capable of independent reasoning and action, impacting customer engagement and decision-making. The development of AI factories and sovereign AI infrastructure, as highlighted by industry leaders, emphasizes the need for control, data protection, and compliance with ethical standards. A production-ready GenAI solution, such as the AML Agent Assistant co-developed by Dataiku and NVIDIA, showcases how agentic systems can aid human investigators in workflows like anti-money laundering without compromising auditability. This system, leveraging NVIDIA's foundation models and Dataiku's orchestration tools, automates data gathering and risk assessment while maintaining human oversight, offering a scalable, compliant architecture suitable for various financial applications like fraud detection and Know Your Customer processes.