Ampega and d-fine jointly conducted a proof of concept to explore how agentic AI can effectively support the validation of risk and valuation models in asset management.

The results are promising: by deeply integrating multi-agent systems, “Talk to Your Data” capabilities, and AI-supported analyses, routine tasks can be automated, quantitative assessments accelerated, and validation experts meaningfully supported.

What is particularly encouraging is that the value of agentic AI goes beyond automating existing analyses; it also supports the development of new analyses, scripts, and data evaluations, enabling additional high-value deep dives.

Strong results were observed in “Talk to Your Data” applications, statistical tasks, quantitative analyses, and the qualitative assessment and documentation of validation results.

At the same time, human experts remain firmly in the loop: validators continue to review, steer, and take responsibility for the results, while AI creates more room to focus on key economic questions and quality assurance.

Our conclusion: the PoC impressively demonstrates that, when properly connected with data infrastructure, technical systems, and governance, agentic AI can unlock substantial potential for efficiency gains, scalability, and quality improvements in model validation.