Premier Member Editorial: Agentic AI is Rewriting What Mortgage Lending Looks Like
Casey Williams is the General Manager of Global Mortgage at nCino, where he leverages his extensive expertise to drive innovative solutions in the financial services industry. Beginning his nCino career at SimpleNexus, now nCino’s Mortgage Suite, Casey has worked with both the product and the customers, impacting over 2 million end users throughout the mortgage experience. With a wealth of experience in both customer success and mortgage, Casey is instrumental in helping nCino deliver cutting-edge technology that empowers financial institutions to streamline their processes and deliver superior customer experiences.
Will Jung serves as nCino’s Chief Technology Officer, responsible for leading the company’s engineering, architecture, and AI platform strategy across a global organization. With a career built equally in banking and technology, he translates deep financial services expertise into intelligent solutions for the agentic banking era. Prior to becoming CTO, Will served as nCino’s General Manager of Product in Australia, and before joining nCino spent 15 years at Macquarie Group leading product and technology across retail banking, business banking, and client engagement.
For years, automation followed a script. You mapped the steps, the system ran them, and anything that fell outside the map landed back in someone’s queue. It handled the predictable middle and left the hard cases to people. Each generation got a little faster, but the shape never changed: automation executed and people decided.
Agentic AI works the other way around. You give it the outcome and it figures out how to get there, including for the files that never fit the template, the exceptions where most of a lender’s cost and time actually live. The machine isn’t running your steps and delivering answers anymore; it’s delivering outcomes.
That’s the part most lenders haven’t caught up to. The moment the technology starts deciding, the question stops being how fast can we automate and becomes which decisions do we hand over, and which ones does a person still have to put their name on. Letting an agent run is the easy part. Knowing where to let it run strategically is the whole game and most lenders are debating the wrong half.
The shift is real, but it only pays off if the AI is built around the people who still own the decisions that matter. And that starts with recognizing they don’t all do the same job.
Not every seat needs the same AI
Much of the AI being sold to lenders right now is task oriented. It automates a discrete action and calls it progress. Role-based AI starts somewhere else: who is this person, what are they accountable for, and what would actually make them better at it?
A loan officer is building relationships, managing expectations and shepherding a borrower through one of the most emotional purchases of their life. An underwriter is weighing risk, checking a file against GSE guidelines, and deciding whether a condition can clear or needs a second look. Those aren’t two flavors of the same job, and they shouldn’t get the same AI. The intelligence a loan officer needs — where a file is stuck, what the borrower needs to hear next — looks nothing like what an underwriter needs, which is closer to a clean read on whether the numbers, the documentation and the guidelines line up.
Lenders aren’t interchangeable either. A shop built on purchase business in one market has different priorities than one chasing refinance volume or a particular borrower segment. A retail branch runs on different rhythms than a consumer-direct or broker channel. In an industry where nearly everyone is selling the same rate, how a lender shapes that human touch may be the thing that sets them apart.
There’s also a tougher reason role matters. Mortgage lending is heavily regulated, and the question of what is allowed to do what, take an application, review an AUS finding, clear a title or appraisal condition, make the credit call, deserves the same rigor whether the approver is a person or an agent. That’s the judgment question again, in its most concrete form. Get the lanes right and you have an answer when an examiner asks who made the decision, and on what basis. Get them wrong and it won’t matter how fast the agent was.
The relationship starts earlier and lasts longer than the loan
Lenders who think their job begins at application and ends at closing are leaving the most valuable part of the relationship on the table. A borrower starts forming the intent to buy long before they fill out a form, and that fuzzy early stretch is where an agent can do its most useful work of helping someone figure out whether they’re buying a first home, an investment property, upsizing or downsizing. That kind of guidance used to require a trusted advisor and a lot of time, which put it well outside the process as most lenders define it.
It doesn’t end at the closing table either. The loan officer who funded a mortgage 18 months ago has a client whose rate picture has since changed. An agent can watch that portfolio continuously, flag the borrower who now qualifies for better terms, and prompt the officer to reach out before someone else does. The mortgage may not be the most profitable thing a lender sells, but it’s the front door to everything behind it: home equity, small business lending, investment accounts. An agent that can see the whole relationship makes the whole relationship more valuable.
Agents will sort the loan officers, not replace them
Cost-to-originate comes up at every industry conference, and agentic AI will change that math, though not the way most people expect.
The loan officer who generates business by being relentlessly present in the community, whose referral network runs from real estate agents to the PTA, is creating value no agent will touch. That advantage is human in a way you wouldn’t want to automate even if you could. But the loan officer whose value is essentially transactional, waiting on files, chasing conditions, moving paper between systems, is doing work an agent will simply do better. The lenders who see the difference early will aim AI where it actually changes the outcome, and double down on the people who do what no agent can.
The cost of waiting
The lenders in the strongest position a year from now won’t be the ones with the most ambitious roadmap. They’ll be the ones who picked one use case, shipped it, and learned from it — a single workflow, a single agent experience, enough to start closing the gap between what the technology can do and what the organization actually knows how to use.
It sounds obvious. But the number of lenders still adding to the whiteboard without crossing anything off suggests the obvious is harder to act on than it looks. The ones who move now will have spent a year learning the only thing that actually matters in this shift is not how to make the technology run, but where to trust it and where to keep a person’s name on the decision.
That’s the half worth getting right.
(Views expressed in this article do not necessarily reflect policies of the Mortgage Bankers Association, nor do they connote an MBA endorsement of a specific company, product or service. MBA NewsLink welcomes submissions from member firms. Inquiries can be sent to Editor Michael Tucker or Editorial Manager Anneliese Mahoney.)
