Financial Services Industry (FSI) Cohort

AI is moving fast. Getting it into production is the hard part. Financial institutions do not have a shortage of AI tools, models, pilots, or ideas. The bigger challenge is building the operating system around them. Who owns the investment? Who funds it? What has to happen before an AI system reaches production? How should model risk evolve as systems become increasingly autonomous? And ultimately, how do you prove that AI created measurable business value? Michael Upchurch joins the AI Ready RVA Financial Services Industry Cohort for a practical conversation about where AI in financial services actually stands today. Michael has spent more than 25 years working across financial services, analytics, machine learning, strategy, product, governance, risk, and enterprise AI. At Capital One, he directed enterprise machine learning strategy and portfolio management across nine lines of business and helped build the operating framework that shifted a 135-person Center for Machine Learning from engineering-led to product-led delivery. He later led Financial Services and Insurance strategy at Domino Data Lab, working with major banks and insurers on AI adoption, MLOps, and model risk management. Today, Michael is Founder and Principal of Eddington, a Richmond-based working lab focused on moving enterprise AI from strategy into governed production and measurable value. What We’ll Explore Where the industry actually is There is a major difference between experimenting with AI and operating AI in production. We will examine what that gap looks like inside regulated financial institutions. Why the operating model matters Technology is only part of the equation. Ownership, funding, decision rights, delivery standards, governance, adoption, and production criteria determine whether AI investments become enterprise capabilities. Model risk in an agentic world Traditional model-validation approaches were designed around statistical and machine-learning systems. AI agents introduce