Fintech Hunting

The episode contains solid operational and strategic insights about AI implementation in mortgage lending, vendor selection, and product mix strategy. However, much of the content consists of name-dropping tools and vendors without deeply exploring *why* they work or providing quantifiable ROI metrics beyond broad claims. The guest offers useful frameworks (sales-first AI implementation, portfolio-first approach for compliance, build vs. buy shop mentality) but these are moderately novel rather than exceptional.

“we are working our AI strategy and sales first and I think that's the right path, um, to do it”

“if you load up new opportunities and new business now the fulfillment operations teams are busy, so then you come and give the lift with AI behind that”

The guest presents some contrarian positioning (e.g., waiting on AI adoption may not hurt you; local lenders can compete with tech) but largely recycles industry talking points about trigger leads, rate sensitivity, non-QM strategy, and vendor consolidation. The 'sales-first' AI approach is sensible but not particularly novel. The episode lacks first-principles thinking or genuinely unexpected claims that would surprise an operator familiar with mortgage market dynamics.

“by the time you do, it'll be the best version of it it's ever been, and you may leapfrog your competition then”

“the technology is helping people fight that guerrilla tactic warfare and then also they're local”

Kevin Perennial is a CLO and partner at PRMG with clear operator credibility and multi-year experience deploying AI and technology at scale in a real lending business. He demonstrates domain expertise across macroeconomics, operations, and technology. However, he is primarily a service provider/vendor ecosystem player rather than someone who has built a major disruptive lending platform from scratch, which limits caliber to 'very good' rather than 'elite.'

“Chief Lending Officer and partner at PRMG”

“We have an AI, um calling assistant that we've rolled out two months ago and um, had our second loan fund uh, this week. Uh, and yeah, so it's, it's making 2,500 calls a week”

The guest provides concrete details on his own deployments (2,500 AI calls per week, 30% through Lone Express, vendor rationalization from 252 to fewer). He names specific vendors and tools (Lotusoft, Capacity, Sella, Halcyon, Grove's IQ). However, he provides minimal quantified outcomes - no ROI figures, time savings metrics, cost reductions, or customer impact data. Many claims remain illustrative rather than evidenced with hard numbers.

“We have an AI, um calling assistant that we've rolled out two months ago and um, had our second loan fund uh, this week. Uh, and yeah, so it's, it's making 2,500 calls a week on just 3,500 contacts”

“We had 252 vendors and we went through every single one of them”

The host asks sensible, strategic questions and shows genuine curiosity about vendor vetting and AI ROI. However, the conversation lacks sharp pushback or challenging follow-ups. When the guest makes broad claims ("AI everywhere," "huge productivity boost"), the host doesn't ask for proof, timelines, or specific metrics. The discussion of 'waiting for rates' is superficial and the host doesn't prod on whether that strategy is actually viable or self-defeating. The tone is collegial but soft.

“How do you figure out what should plugged in and what is not worth your time?”

“What are some of the solutions you're most excited about?”