The U.S. mortgage industry was not built with every borrower in mind.
Over the past few months, a growing number of essays and firsthand accounts have surfaced the same issue from different angles: people with disabilities routinely face barriers when trying to access a mortgage. Not because the industry sets out to exclude them — outright discrimination is rarely the intent — but because the process itself was designed around a standardized borrower. One who reads at a certain pace, processes dense financial language without friction, and moves through a rigid sequence of forms, disclosures, and deadlines without needing anything different along the way.
That design assumption breaks down quickly in practice. Cognitive disabilities, learning differences, visual or hearing impairments, and even situational factors like stress or unfamiliarity with financial systems all change how someone needs to receive and act on information. A borrower with a cognitive disability may need concepts broken into smaller steps, repeated in different ways, or explained without jargon. Someone with a visual impairment may need documents in a format their screen reader can actually parse, not a scanned PDF. Someone processing a lot of information at once may simply need more time, and a process that doesn't penalize them for taking it.
None of this is exotic. It's the ordinary range of how people think, read, and make decisions. But a mortgage process built around a single "standard" path treats that range as an edge case, and edge cases are, by definition, the first thing to be underserved when a system optimizes for the average.
The result is a familiar pattern: people who are otherwise qualified borrowers get filtered out not by their creditworthiness, but by their ability to navigate a process that wasn't built with them in mind. That's not a lending problem. It's an access problem — and access problems are, in principle, solvable with better design.
An attempt at a different layer
For the past few months, I've been building an AI assistant in my spare time to explore one possible answer to that problem.
The premise is simple: create a layer between the borrower and the mortgage process — one that adapts to the individual rather than asking the individual to adapt to it. Concretely, that means understanding a borrower's level of comprehension, adjusting the pace and format of information accordingly, and providing the right kind of assistance at the right moment, throughout the entire journey rather than at a single point of friction.
I want to be clear about what this is not. It's not an attempt to replace human interaction, and it's not a bet that technology alone can fix a structural problem. Loan officers, housing counselors, and support staff play a role that software shouldn't try to substitute. What I'm trying to do is narrower: use technology to make the process more natural, more efficient, and, above all, more inclusive for the people currently being left behind by it.
Where I think the starting point should be
The starting point should always be the human layer. Technology comes after, not before.
People will always require different levels of guidance, and a truly inclusive process doesn't expect everyone to fit into the same workflow. Instead, the workflow should adapt to people — regardless of their cognitive abilities, physical abilities, or any other support needs. Designing for that reality isn't a matter of adding an accessibility feature on top of an existing product. It changes the starting assumptions: comprehension isn't uniform, pace isn't uniform, and the "right" way to present a disclosure or a decision point isn't a single fixed format.
That's a harder problem than it sounds, and I don't think it has a finished answer yet. But it feels like the right problem to be working on.
If you work in the mortgage industry, in accessibility, or in AI, and this resonates with what you've seen — I'd like to hear about it.