The LO the Industry Is About to Lose

In my last field report, I wrote about the AI tools that launched at ICE Experience 2026. Automated conditioning engines. Borrower engagement autopilots. Virtual economists. The biggest names in AI sitting on stage at a mortgage conference.

But the piece I couldn't stop thinking about wasn't about the tools. It was about the people.

If AI handles the transactional work that most loan officers spend their day doing, what happens to the loan officers whose entire value was the transactional work?

Nobody at the conference said it out loud. So I will.

Not every loan officer is at risk. But a specific type is. And it's not defined by age, company, or channel. It's defined by what they actually do all day.

The order taker.

This is the LO whose value proposition is: "I can take your application and push it through the system." That's the job. Take the 1003, upload the docs, chase the conditions, send the status updates.

Every single one of those tasks is being automated. Not in five years. Now. Blend's autopilot reviews documents and issues follow-ups around the clock. Ocrolus generates conditions and matches them to paperwork without a human touching them. The entire workflow that justified this role is being rebuilt without the role in it.

It doesn't matter if this LO is at a bank, a retail lender, or a brokerage. If the only thing between them and a borrower is a loan application, that gap is closing fast.

The non-adapting veteran.

This is not about age. Some of the best loan officers I know are in their 50s and 60s and they're thriving because their value was never the technology. It was the relationships. The referral networks built over decades. The reputation in their community.

The ones at risk are the veterans who've been coasting on a shrinking book of past clients. Doing two or three refi deals a month. Haven't originated a purchase loan in years. When rates don't cooperate and AI handles the simple stuff, there's no chair left.

The refi wave isn't coming back the way it did before. And the transactional skills that carried someone through 2020 and 2021 aren't transferable to a purchase market that demands advisory-level work.

The call center LO.

The loan officers at big retail lenders who sit in a cube, take inbound internet leads, and race to lock a rate. This is the assembly line model. High volume, low touch, speed over substance.

AI and automation can handle that entire funnel. Rocket already proved the model years ago. Now every large lender is building it. These LOs are essentially doing quality assurance on a process that won't need quality assurance much longer.

When the AI can qualify a borrower, pull the docs, check compliance, and issue a pre-approval letter without human intervention, the only question is how many humans you still need in the middle. The answer is fewer than today.

Now, here's the part where I push back on my own argument.

Not all of these roles disappear as cleanly as the thesis suggests. The call center model doesn't need every LO to be individually profitable. It needs the system to be profitable. AI might reduce headcount at those shops, but it won't eliminate the model. It'll optimize it. Some order-takers keep their seats because the machine around them gets cheaper to run. The assembly line doesn't shut down. It just needs fewer people on the line.

And the timeline matters. The median LO in America isn't reading this playbook. They're not on LinkedIn debating automation. They're doing two or three loans a month at a community bank and their biggest technology complaint is that their LOS is slow. The disruption timeline for that person is measured in years, not quarters. Adoption in mortgage has always been slow. That hasn't changed just because the technology got faster.

I still believe the direction is right. But the speed is uneven. And pretending otherwise would be the kind of oversimplification this playbook is supposed to push against.

The mortgage industry has roughly 300,000 licensed loan officers. A significant portion of them close one to three loans a month doing mostly transactional work.

Here's the math that keeps me up at night.

If AI gives a top producer three extra hours a day, that LO goes from 8 loans a month to 12. Maybe 15. Their capacity expands because the overhead shrinks.

But I need to be honest about the assumptions in that math. The 8-to-15 jump assumes the bottleneck is time. For a lot of top producers, it's not. It's referral flow. It's market conditions. It's inventory. Giving someone three extra hours doesn't create three extra deals if there aren't three extra houses to sell. The time-to-volume math only holds in a purchase market with enough supply. Right now, that's not most markets.

The other side of the equation is more straightforward. If AI eliminates the transactional tasks that justified a marginal LO's existence, that LO goes from 2 loans a month to zero. Not because they got fired. Because they have nothing left to offer that the borrower can't get from the platform directly.

The overhead of licensing, compliance, and E&O insurance doesn't make sense for someone closing 15 loans a year when an AI-augmented LO down the hall is closing significantly more.

The industry is heading toward fewer loan officers doing more volume each. Not because companies will lay people off in dramatic waves. Because the bottom tier will stop being economically viable on its own. That part of the math I'm confident in. The top-end expansion depends on factors AI can't control.

Brokers are in an interesting position. They have a structural advantage and a structural vulnerability at the same time.

The advantage. Brokers shop wholesale rates across dozens of lenders. AI makes that comparison faster, more accurate, and more transparent. A tech-savvy broker with AI tools can deliver better pricing, faster, with less overhead than a retail LO locked into one lender's rate sheet. The value proposition of "I find you the best deal" gets stronger with AI, not weaker.

The vulnerability. The broker model depends on the individual producer. There's no corporate brand carrying you. No national ad spend. No app with your name on it. If a broker doesn't adapt to AI-powered workflows, there's no bank infrastructure to fall back on. They just lose.

And the barrier to entry for brokers is low enough that new, tech-native brokers will enter the market and immediately outperform the ones who've been doing it the old way for 20 years. Same licenses, same wholesale access, but completely different operating system.

My prediction. The broker channel grows in market share. It already has been. But the number of individual brokers shrinks. Fewer brokers, each doing more volume. The survivors will be the ones who combine wholesale access with AI efficiency and genuine relationship skill. The rest get absorbed or retire.

There's a cost to that prediction worth naming. If the industry consolidates around AI-powered super-producers, the path for a new LO to break in gets harder, not easier. Fewer brokers doing more volume also means fewer entry points for the next generation of originators. That's a workforce development problem nobody in the AI conversation is talking about. The same efficiency that rewards the survivors raises the floor for everyone trying to get started.

Here's what AI can't do.

It can't sit across from a first-generation homebuyer and make them feel safe. It can't explain what earnest money is in a way that doesn't make someone feel stupid for asking. It can't switch to Spanish mid-sentence because that's the language the borrower's parents speak and you can feel the tension drop when you do.

It can't call a realtor partner after they lost a deal, walk through what went wrong, and build a plan for next time. That's not a transaction. That's a relationship. And relationships generate the next five referrals.

It can't structure a creative deal on a self-employed borrower with two years of declining income but strong bank deposits. It can't layer a down payment assistance program for a borrower on a work visa who doesn't know the process and doesn't speak English. These files need a human who knows the guidelines cold and can think within them.

It can't coach a junior LO through the moment where a 4506-C stops being confusing and starts making sense. AI can generate the training materials. It can't deliver the moment of understanding.

The LO who survives and thrives is part advisor, part therapist, part deal architect. The one a borrower calls not because they have to, but because they want to.

The industry isn't losing "loan officers" as a category. It's losing a type of loan officer. The transactional, non-advisory, non-technical LO who adds no value beyond what a form and a processor can handle.

That role is going away whether the person holding it is 30 or 60.

This isn't something that happens to the industry. It's something that's already happening inside it. The LOs who see it are building new skills. The ones who don't are wondering why their pipeline feels different and blaming the market.

The market isn't the problem. The market is just the thing that makes the shift visible.

I don't write this to scare anyone. I write it because the people who read this playbook deserve the honest version. And the honest version includes the parts that complicate my own argument.

The shift is real. The direction is clear. But the pace is uneven, the variables are messier than any single framework can capture, and the macro forces driving this industry (rates, inventory, regulation) don't care about AI timelines. Anyone who tells you they know exactly how fast this plays out is selling something.

What I do know: AI is going to make the best loan officers in this business dramatically better. It's going to give them more time for the work that matters, more capacity for complex files, more bandwidth for the relationships that drive referrals.

And it's going to make the transactional parts of this job unnecessary.

If your value is in the transaction, start building value somewhere else. Now. Not next quarter. Not when your company rolls out a training program. Now.

If your value is in the advisory work, the relationship work, the complex problem-solving that no platform can replicate, then AI isn't your threat. It's your leverage.

The runway is still open. But it's shorter than most people think. Even if nobody can tell you the exact length.

First in. Documenting everything. Building the frameworks for what comes next.

\- Joshua Rios