Why AI matters at this scale

Shay Hensley—Powered by Union Savings Bank operates as TopChoice Mortgage, a residential mortgage brokerage based in Mason, Ohio. Founded in 2019 and employing 1,001–5,000 people, the company facilitates mortgage loans, connecting borrowers with lenders. Its online platform, topchoice.mortgage, suggests a digital-forward approach in a traditionally paper-intensive industry. As a mid-market player, the company has sufficient transaction volume to generate valuable data but faces pressure to streamline operations and enhance customer experience in a competitive, rate-sensitive market. AI adoption is crucial for scaling efficiently, reducing manual errors, and gaining a competitive edge through personalized, faster service.

Concrete AI Opportunities with ROI

1. Automated Document Processing with NLP: Mortgage underwriting involves manually reviewing hundreds of pages per application. Implementing AI-powered optical character recognition (OCR) and natural language processing (NLP) can automatically extract and validate data from pay stubs, W-2s, and bank statements. This reduces processing time from several days to hours, cuts labor costs by up to 70% on repetitive tasks, and minimizes human error, directly boosting underwriter productivity and borrower satisfaction.

2. Predictive Analytics for Lead Scoring and Default Risk: By applying machine learning to historical application data and online behavior, the company can predict which leads are most likely to close and which loans might default. This allows loan officers to prioritize high-intent borrowers, improving conversion rates by an estimated 15–20%. Simultaneously, early default risk flags enable proactive mitigation, potentially reducing bad debt provisions.

3. AI-Driven Dynamic Pricing and Recommendation Engines: Integrating AI models that analyze real-time market rates, borrower credit profiles, and competitor offerings can enable dynamic, personalized mortgage rate quotes. This optimizes profit margins while remaining competitive, potentially increasing revenue per loan by 1–2%. A recommendation engine can also suggest suitable loan products to borrowers, enhancing cross-sell opportunities.

Deployment Risks for a 1,001–5,000 Employee Company

At this size band, the company has more resources than a small startup but faces integration complexity across potentially siloed departments (sales, underwriting, compliance). Key risks include: Data Silos and Quality: Historical data may be fragmented across legacy systems, requiring upfront investment in data consolidation and cleansing for reliable AI models. Change Management: Scaling AI requires training hundreds of loan officers and underwriters on new tools, with potential resistance to altered workflows. A phased pilot program is essential. Regulatory and Explainability Hurdles: Mortgage lending is highly regulated. AI models used for credit decisions must be explainable to comply with fair lending laws (e.g., ECOA). Black-box models pose significant compliance risks, necessitating investments in interpretable AI or human-in-the-loop systems. Vendor Lock-in: Relying on third-party AI SaaS platforms can lead to dependency; building internal AI competency is a strategic but costly countermeasure.