Mortgage AI From an Operator’s Perspective ft. Jason Harris

This is a summary of an episode of Pioneers, an educational podcast on AI led by our founder. Join 3,700+ business leaders and AI enthusiasts and be the first to know when new episodes go live. Subscribe to our newsletter here.

TL;DR:

• Mortgage lending has become a commodity business, making automation and AI the primary sources of competitive advantage.
• AI agents are first succeeding in low-risk, repeatable operational workflows hidden from customer view.
• Lenders do not need 100 percent automation to win. Offloading 80 percent of work already creates massive leverage.
• Agentic AI helps lenders scale without hiring and firing through economic cycles.
• Institutions that delay AI adoption risk being structurally uncompetitive as margins continue to compress.

Before we dive into the key takeaways from this episode, be sure to catch the full episode here:

Meet Jason - Strategic Initiatives Manager at American Portfolio Mortgage

Jason Harris is Strategic Initiatives Manager at American Portfolio Mortgage and has spent over two decades in the mortgage industry. His career spans the evolution from carbon copy paperwork and phone-based verifications to automated underwriting and modern digital workflows.

Jason has seen multiple waves of automation, from offshore process outsourcing to RPA and now agentic AI. His perspective is grounded in operational reality, regulatory complexity, and thin industry margins.

At American Portfolio Mortgage, Jason focuses on how lenders can use AI agents to eliminate mechanical work while preserving the human side of lending. He believes automation should free professionals to focus on advising borrowers, improving service quality, and scaling sustainably without cyclical hiring and layoffs.

Where AI Agents Create Immediate Value Today

Jason sees AI agents delivering results first in stable, repeatable workflows hidden from borrowers. “Whatever is the most stable repeatable process is generally the safest,” he explains.

One example is automated disclosures, a task that previously required large teams and minutes of manual effort per loan. “They went from having a staff of ten people to just one person,” Jason says, noting that humans now handle only exceptions. These workflows do not break compliance and dramatically reduce labor costs.

“It’s not going to do 100 percent of anything today,” Jason clarifies. “But can it do 80 percent? Absolutely.” That mindset unlocks immediate operational leverage without risking customer experience.

“If you’re not looking at it, you’re going out of business” — Jason Harris

The 80 Percent Rule for Automation in Lending

Jason frames AI agents as managers, not replacements. “If you can delegate 80 percent of something, delegate it,” he says. In lending, that 80 percent includes income analysis, document review, disclosures, and prequalification work.

Humans retain oversight and decision authority. “Let it do all the thinking and let me review its work before I communicate the decision,” Jason explains. Over time, accuracy improves. “Eighty becomes ninety, becomes ninety-five,” he adds.

This gradual progression aligns with regulatory caution while still delivering meaningful gains. Jason emphasizes that perfection is not required for impact.

“Machines don’t make mistakes like humans make mistakes,” he notes, highlighting error reduction as a major source of value.

“In the future, there are going to be two types of companies: ones with great AI and the ones that closed.” — Jason Harris

Scaling Without Hiring in a Cyclical Industry

Mortgage lending is highly cyclical, and Jason describes the human cost of constant hiring and layoffs. “It feels awful to hand somebody a pink slip,” he says.

AI agents offer an alternative. “It’s freeing up labor capacity,” Jason explains. Lenders can retain teams, absorb volume surges, and improve service without adding headcount.

“That little incremental savings is massive in an industry that only makes ten basis points,” he notes. Instead of cutting staff, organizations can redirect people to customer-facing work, marketing, and advisory roles.

“You can handle a limited surge of business without hiring a new person,” Jason says, describing a more humane and sustainable operating model.

“Having agents in the background speeding up the process is your differentiator.” — Ankur Patel

Build, Buy, or Partner in an Agentic Future

Jason believes most lenders cannot rely on legacy loan origination systems for AI. “They don’t have agentic anything,” he says.

Building internally requires deep pockets that only the largest lenders possess. “The big players are putting hundreds of millions into this,” Jason explains. For most organizations, partnering or building on specialized platforms creates flexibility and differentiation.

“If everyone has the same tools, how are you going to win?” he asks. Competitive advantage comes from customizing workflows and ideas, not from using identical systems. “Now it becomes who has the best ideas and who can move fastest,” Jason says, framing AI as a democratizing force for smaller lenders.

Would you like to learn more about agentic AI for the mortgage industry? Check out this episode on how AI can solve mortgage crisis with Jen Peachman.