Borrowers have a new agent. Is your organization built for it?

AI is shifting negotiating power from lenders to borrowers at scale. The institutions that restructure their workflows and tools to meet that reality will grow. The ones that don’t will feel it in pull-through rates before they understand why.

For most of mortgage’s recent history, information asymmetry was a structural advantage. Most borrowers didn’t know what they didn’t know about pricing, programs, or what their credit profile actually entitled them to. That advantage is eroding faster than most executives have accounted for in their strategic plans.

Generative AI has given consumers a research partner that is always available and increasingly fluent in mortgage. A JD Power survey from July 2025 found that 20% of consumers have already used AI for loan or mortgage research—and 59% are using it at least occasionally for banking and financial services. A Menlo Ventures study shows adoption growing across all generations, with particularly strong use among households earning more than $100K a year. These are not marginal borrowers. They are your highest-value prospects, arriving at the point of sale better prepared than ever before.

The institutions watching this as a cultural curiosity are making a strategic error. This is a structural shift in borrower behavior with significant competitive implications.

What happened to other industries is instructive

Zillow didn’t eliminate real estate brokerages. It eliminated the version of a brokerage whose primary value was information access. The firms that thrived repositioned around execution, interpretation, and service complexity. The ones that failed to pivot lost ground steadily and largely didn’t understand why until the migration was well underway.

Online travel agencies did the same to the legacy travel industry. Comparison shopping became default behavior almost overnight. Standard transactions got commoditized, and firms that had built their business on volume without differentiation got squeezed.

Mortgage faces an analogous reset, with one important distinction – information asymmetry has historically been a larger share of the value proposition here than in either of those industries. That makes the exposure greater and the need to adapt more urgent.

The silent attrition problem

The most significant near-term risk isn’t the borrower who actively pushes back on an offer. It’s the borrower who leaves without a word.

When a borrower arrives having modeled their scenarios with AI — knowing what their credit score entitles them to, what rates are available, and what a fair deal looks like —they are evaluating the loan officer and their offer against a baseline they brought with them. If they sense that they could be doing better, they will simply move on to the next lender.

We are starting to see this in client production data – loan officers who proactively produce a credit optimization plan close at materially higher rates than those who don’t. The explanation is not that credit optimization improves every outcome. It’s that borrowers who don’t receive a plan are more likely to shop—and the borrowers most likely to shop are also the most creditworthy, the most financially sophisticated, and the most valuable.

The deals being lost are not appearing in your pipeline as lost deals. They are appearing as top of funnel prospects that quietly slip away. That’s what makes the problem easy to underestimate.

Stakes are high for borrowers . . . and your pipeline

The financial stakes for a borrower in a relatively high interest rate environment are not insignificant. On a $500,000 purchase with 10% down, the difference between a 689 and a 740 credit score represents $301 less per month, $3,612 per year, nearly $39,512 over a decade. An AI-informed borrower has likely done their homework.  If your loan officers are not producing plans that help borrowers reach a target score, you can be sure that their competition will be.

Scale the above situation across your annual pipeline. Even modest improvements in pipeline pull-through among mid-to-high credit borrowers will compound quickly. On the other hand, steady attrition among your most creditworthy prospects will have portfolio-level consequences that could be difficult to reverse as referral relationships shift to the competition.

The limits of AI and a potential compliance exposure 

There is a version of this story where lenders conclude that borrowers doing their own AI research is simply a new normal to accommodate. That conclusion misses the more actionable insight.

Generative AI is probabilistic and has real limitations when it comes to specific recommendations. While it is helping to educate borrowers on credit basics and model mortgage pricing scenarios, it is not equipped (or designed) to give detailed guidance.  To be clear, generative AI cannot analyze a specific borrower’s file and produce account-specific guidance with a quantified probability of reaching a target score within a defined timeline.

More importantly, the moment a loan officer uploads a credit report to a general-purpose AI system, your institution has a compliance and liability exposure that many legal and risk teams have not yet fully mapped. Sensitive personal financial data entering a large language model without clear policies creates regulatory risk that is difficult to quantify and even harder to remediate after the fact.

The opportunity for mortgage lenders lies in the gap between what AI can tell a borrower in general terms and implementing a secure, purpose-built and highly accurate platform that can be used to produce clear steps to achieving a target score. 

The strategic question for leadership

The question for C-suite leaders is not whether borrower sophistication is increasing. It is. The question is whether your organization’s systems, training, and workflows are built for a more empowered borrower that is looking for precision and execution.

Data consistently shows that roughly 70% of all borrowers can improve their score by at least 20 points in 30 days. That is not a niche population—it is the majority of your pipeline. Treating optimization as a reactive intervention for borrowers who don’t qualify, rather than a standard offering at every client engagement, is a structural inefficiency that will compound over time.

The institutions that will grow in this environment are those that invest in the infrastructure to deliver certainty: documented improvement plans, quantified timelines, tracked milestones, and a borrower experience that delivers on their expectations. That infrastructure will deliver returns through improved pull through rates as your team significantly reduces comparison shopping —borrowers that are immediately presented with specific, credible plans from your loan officers have little incentive to shop.

The window for differentiation is open, but not indefinitely

The borrowers coming through your door next year will be better informed than the ones your team is closing now. Lenders who build the workflows to meet that reality today will establish a reputation for execution, expand referral relationships and form a durable competitive advantage.

The ones who simply train loan officers to sound more informed will find that borrowers are not easily impressed by fluency in concepts they already understand. What the AI informed borrower is looking for is execution. The institutions that meet borrowers where they are and deliver will earn their business. The ones that don’t will lose it quietly, as their pipelines migrate to those focused on execution.

Mike Darne is the VP of Marketing at CreditExpert.
This column does not necessarily reflect the opinion of HousingWire’s editorial department and its owners. To contact the editor responsible for this piece: [email protected].