AI Agent Operational Lift for Loan One, A Division Of The Union Bank Company in Gahanna, Ohio

Why AI matters at this scale

Loan One, a division of The Union Bank Company, is a mid-market mortgage lender headquartered in Gahanna, Ohio, with 201-500 employees and roots dating back to 1904. The firm originates residential mortgages through a blend of retail and broker channels, relying heavily on manual underwriting, document collection, and compliance checks. At this size—large enough to have meaningful data but small enough to lack massive IT budgets—AI offers a disproportionate advantage: automating high-volume, repetitive tasks that currently consume underwriters and processors, while improving consistency and speed in a highly regulated environment.

Mid-market mortgage lenders face intense pressure from both larger digital-first competitors and smaller, agile brokers. Margins are thin, and loan officer productivity directly drives profitability. AI can compress cycle times, reduce fallout, and lower cost-per-loan by 15-25%, making it a strategic lever for growth without proportional headcount increases. With a likely tech stack centered on Encompass or Calyx LOS, Loan One already captures structured and unstructured data that can fuel machine learning models with minimal integration friction.

Three concrete AI opportunities with ROI framing

1. Intelligent document processing and data extraction

Mortgage origination still drowns in paper and PDFs—pay stubs, W-2s, bank statements, tax returns. An AI-powered document intelligence layer can classify, extract, and validate data from these documents in seconds, feeding directly into the LOS. For a firm processing 3,000-5,000 loans annually, this can save 20-30 minutes per file, translating to 1,000-2,500 hours reclaimed per year. ROI comes from faster underwriting turn times, fewer condition reviews, and reduced manual keying errors that cause costly re-disclosures.

2. AI-assisted underwriting and compliance screening

Rather than replacing underwriters, AI can pre-screen files against investor guidelines and regulatory requirements (TRID, ECOA, state-specific rules). The system flags missing documents, income calculation discrepancies, or potential fair lending red flags before a human touches the file. This reduces the back-and-forth that plagues mortgage processing and lowers the risk of compliance violations. Even a 10% reduction in underwriting touch time can increase loan officer capacity by 5-8 loans per month across the team.

3. Predictive pipeline management and borrower engagement

Machine learning models trained on historical pipeline data can predict which loans are likely to close, which need intervention, and when borrowers are at risk of disengaging. Paired with an AI chatbot for status updates and document nudges, Loan One can improve pull-through rates by 5-10 percentage points. For a $500M annual origination volume, that represents $25-50M in additional closed loans with minimal incremental cost.

Deployment risks specific to this size band

Mid-market lenders face unique AI adoption risks. First, data quality and fragmentation: loan files may be spread across LOS, CRM, and shared drives, requiring upfront data engineering. Second, regulatory scrutiny: CFPB and state examiners expect explainable, non-discriminatory models—black-box AI is unacceptable. Third, change management: loan officers and underwriters may resist tools they perceive as threatening their expertise or commissions. Mitigation requires phased rollouts, transparent model logic, and clear communication that AI handles the grind, not the judgment. Finally, vendor lock-in: many mortgage-specific AI tools are bundled with LOS upgrades; Loan One should prioritize API-first, interoperable solutions to maintain flexibility.