AI Risk Management Agents for an Australian Bank

An Australian bank needed to enhance its risk management capabilities to meet complex regulatory demands. Traditional methods were resource-intensive and difficult to scale. A technology partner was engaged by the bank to develop a sophisticated application using AI agents and subsequently subcontracted MXA to design and build the core AI components. The critical challenge was developing advanced AI capabilities while meeting stringent requirements for data security, transparency, explainability, and bias mitigation.

The bank faced the challenge of enhancing risk management in an increasingly complex regulatory environment. The objective was to increase both efficiency and breadth of coverage. Traditional methods were proving resource-intensive and difficult to scale, and any AI solution needed to meet the bank's rigorous compliance requirements for security, transparency, explainability, and anti-bias.

MXA designed a robust, scalable agent architecture prioritising security, modularity, and compliance. We developed the back-end software and integrated it into the core application. We established comprehensive LLMOps pipelines to manage the entire AI model lifecycle.

Rigorous automated testing protocols were implemented, focusing specifically on validating explainability and anti-bias requirements. MXA managed deployment into the bank's production environment and continues to maintain the AI agents under a SaaS model.

The AI-agent system is now fully operational. The project significantly increased the efficiency and coverage of the bank's risk management activities. A key achievement was successfully navigating the bank's rigorous compliance gateways, meeting all requirements for security, transparency, and anti-bias. MXA continues to maintain and enhance the AI agents, ensuring the system remains optimised and adaptive to new risks.