The AI Spending Paradox in Lending

The AI race in lending is real. The returns are not — at least not yet, and not for most. The numbers tell a consistent story. KPMG's 2026 Banking Technology Survey found that 80% of banking executives expect AI to significantly disrupt their business within three to five years, and their Q2 2026 AI Pulse survey put average projected AI investment for banking organizations at roughly $170 million over the next 12 months. Yet survey after survey finds fewer than one in five financial services executives can point to a single AI initiative that has produced a measurable improvement in a lending outcome. The investment is real. The ambition is real. The connection between them is what's broken. This isn't a technology problem. It's a production problem. Most lending institutions have built AI. Very few have operationalized it. The gap between a model that works in a data scientist's environment and a model that runs in live underwriting decisions — at scale, with governance attached, with the auditability a regulator expects — is where the majority of the industry's AI investment quietly disappears. This invitation-only event brings together senior leaders from across the lending industry —community banks, credit unions, mortgage lenders, private lenders, and specialty finance organizations — for a direct, peer-level conversation about the gap between AI investment and AI results: what's driving it, what it costs, and what it would actually take to close it. Agenda 7:30 AM — Arrival and networking breakfast 8:00 AM — Main session: The AI Spending Paradox in Lending (30 minutes) 8:30 AM — Open Q&A and peer discussion 9:15 AM — Closing remarks and networking 10:00 AM — Close What We'll Cover The main session addresses: The AI productivity paradox in lending: why institutions spending the most on AI are rarely the ones with the most to show for it — and what the data actually says about the gap between AI investment and operational return What "operationalizing AI" means in practice for a lending institution — the specific difference between a model in a lab and a model in a live credit decision The AI lending stack: what the institutions pulling ahead have built — and the organizational and technical decisions that separate scaled AI production from endless piloting The competitive advantage question: when the best lender isn't defined by balance sheet or branch network but by the quality, speed, and governance of its AI enabled credit decisioning, what does that mean for institutions of your size? Open peer discussion: where your organizations are in the AI production journey, what's working, and where the real friction points are. Our Speakers John McCambridge, Global Solutions Director, Financial Services & Insurance, Dataiku John has spent over five years at Dataiku helping lending institutions move from AI experimentation to production-grade deployment. Before Dataiku, he held Senior Vice President roles at Citi leading strategy and business transformation across global commercial and consumer banking in New York and Singapore. At this session, John will speak to what closing the AI productivity gap actually requires, and what separates institutions that have scaled AI in credit decisioning from those still running pilots. Trevor Palmquist, Celestial Systems Trevor has spent two decades working inside private debt and capital markets, raising capital across Canada, the US, and the Middle East, and watching deals move at the speed of whoever had the fastest process. On a recent engagement, Celestial helped a lending client cut throughput time from 30 days to under 36 hours. At this session, Trevor will walk through where AI is actually compressing timelines in production environments, what separates real capability from demo-ware, and how lending institutions can build the internal case for adoption without overhauling their credit culture.