Your Dealership AI Is Only 90% Right. That’s the Problem.

Your Dealership AI Is Only 90% Right. That’s the Problem. AI can make messy data look usable. But AJ McGowan says that gets a lot more dangerous when you expect AI to actually run dealership workflows. AJ, VP of Research and Development at Reynolds and Reynolds, joins Paul, Kyle, and Michael for a conversation about what changes when dealership AI moves beyond chatbots and analysis into truly agentic systems. His argument cuts against an increasingly common assumption: just because AI can figure out relationships between disconnected data doesn’t mean you should make it solve that problem every time. At scale, fuzzy matches become a trust problem—and a 90% answer isn’t good enough when software starts taking action. AJ explains how Reynolds is approaching that challenge with Rey, an AI software agent designed to work across the Reynolds ecosystem, pursue a user’s goal, and pull together information from multiple systems. Today, Rey is being used to answer questions and perform analysis. The bigger step ahead is giving an agent the ability to affect action. Along the way, AJ traces his builder mentality all the way back to coding a game in first grade and building computers for his parents’ friends in sixth grade. It’s a fitting origin story for a conversation about why the next era of dealership technology will still require something surprisingly old-school: understanding the fundamentals well enough to know when AI’s answer can actually be trusted. YOU'LL HEAR ABOUT Why AI can turn a 15-day learning curve into 15 minutes How dealers’ thinking about unified CRM, DMS, sales, service, and inventory data is changing Why messy data becomes more—not less—important with autonomous AI The difference between getting an AI system to 90% and trusting it at 100% How Rey can pursue answers across the Reynolds ecosystem What changes when dealership AI moves from analysis to taking action Why getting your reps in with new technology still matters CHAPTERS 0:00 Intro with Paul J Daly and Michael Cirillo 03:40 AJ’s first software project—and his sixth-grade computer business 06:31 How AI changed the way technologists build 08:13 Compressing 15 days of learning into 15 minutes 11:25 How quickly are dealers actually adopting AI? 12:18 Why unified dealership data matters more with agentic AI 14:29 DAGs, data structure, and whether AI can handle the mess 16:42 Why AJ thinks messy data is still a problem 19:31 The danger of the 90% AI answer 20:29 What Reynolds’ Rey AI can actually do today 23:01 When AI starts taking action inside the dealership 24:31 The next big step for dealership AI—and why reps still matter LINKS AJ McGowan: https://www.linkedin.com/in/ajmcg/ Reynolds and Reynolds: https://www.reyrey.com/ More Auto Collabs: https://daily.asotu.com/auto-collabs Subscribe to the Automotive State of the Union Daily Email: https://daily.asotu.com/