In this episode of Lykken on Lending, Steve Butler, CEO of TRUE, discusses why mortgage cost per loan remains stubbornly high despite years of automation investment and what lenders are missing in their mortgage AI strategies.
Steve makes a clear distinction between buying automation and achieving ROI. He explains that many lenders have invested heavily in tools, yet still feel stuck because they never align automation to granular, task-level workflow changes. TRUE now requires ROI workshops before closing contracts, working directly with processors and underwriters to understand exactly where time is spent and what must change operationally to produce measurable cost savings.
A central theme of the conversation is data quality as the foundation of manufacturing efficiency. Steve argues that most lenders focus too late in the process — underwriting, shipping, or post-close — when the real opportunity lies earlier in loan setup and data validation. When data is incorrect or inconsistent at intake, downstream teams recheck, reopen, and reroute files, creating compounding rework. This “stare-and-compare” culture has become normalized, but it remains one of the biggest drivers of productivity loss.
Steve also addresses why many AI initiatives stall after early success. The issue, he says, is architectural. Vendors often demonstrate strong functional prototypes but lack scalable infrastructure. TRUE invested heavily in a microservices architecture and a common data layer that allows independent services to scale and operate interoperably across the loan lifecycle. That architectural discipline, combined with deep operator-level engagement, has enabled TRUE’s MOS platform to scale rapidly, surpassing 1,000 users.
Looking ahead to 2026, Steve predicts that enterprise-level AI ROI stories will begin to emerge in mortgage. The industry is questioning whether mortgage AI truly delivers financial impact at scale. He believes that lenders who focus on processor productivity first, rather than chasing late-stage optimization, will see 30–50% gains in throughput and real cost-per-loan reductions.
The episode reinforces three major takeaways for lending executives:
- Cost per loan doesn’t drop from automation alone, it drops from workflow redesign aligned to measurable ROI.
- Data quality early in the process determines whether downstream automation succeeds.
- Scalable architecture and interoperability matter as much as AI capability.