Market Moves Sep 21, 2026 56m 36m saved
With Jake Lew, Co-Founder and CEO of Aivre
Aivre was the first appraisal software Fannie Mae and Freddie Mac verified for UAD 3.6, the new residential appraisal report, about six weeks after limited production opened.
Most of the industry has not moved. Jake Lew says his company still meets appraisers who have done no research on the new format and take their first live order as their education. From November 2 the two mortgage enterprises stop accepting the old one.
"But when it hurts that balance sheet, they're going to be the first ones to jump."
Lew graduated from Michigan State in 2013, qualified as an appraiser soon after, and built the software he now sells because he was still typing reports on Sundays. He speaks to people at Fannie Mae and Freddie Mac most weeks, and his platform is one of two carrying the bulk of the 3.6 reports written so far.
The full interview is covered here so you can skip it. 56 minutes of audio, 20 minutes of reading.
Here are the 14 takeaways that matter.
Key Takeaways
A friend asking why he was typing an appraisal on a Sunday is the origin of the company — Lew went home, looked for software that met current tech standards, and found none
The pivot to the new form was a competitive decision, not a product one — on 3.6 every vendor restarts from the same line, and the legacy incumbents lose their installed base The first prototype was built on the old 2.6 form and produced reports in 7 to 12 minutes
Aivre will not let its AI touch the value section — Lew says comp selection is judgment work that a model cannot do
Per-report pricing is the right model for an AI product and the industry refused it, so the cost is forecast and baked into an annual price
Ten platforms are approved and about 15 more are coming, and Lew expects most of them gone within a year Verification is 15 or 16 test cases; he says the separation happens in live production and in support volume
Orders through his platform are close to doubling every month, with two large lenders onboarding and Rocket in pilot
The November 2 date is not moving — Lew talks to the enterprises weekly and says their test cases and numbers are in
A report that filled two pages on the legacy form runs to 15 or 16 pages in 3.6
Without an AI lift the new report can take 15 to 20% longer, which is the whole argument about fees
Lew's prediction is that the last non-agency lenders move by May 2027, using the revision window as a grace period
The appraiser keeps the physical inspection, though he expects hybrid assignments to grow and says the new report was built for them
1. The Sunday That Started It
Mark Verrett opened by reading Lew's LinkedIn back at him: Michigan State in 2013, then straight into appraisal work, then the software company. Lew confirmed it and then went further back than the story the hosts had prepared.
The version Verrett had was about a conversation on a couch. Lew's version starts about fifteen years earlier, in Birmingham, Michigan, where his mother met a woman named Sonia at a coffee shop and the two stayed friends. His mother had spent 28 years at Delta, took a buyout during the pandemic, and went to work as a nanny for Sonia and her husband David's newborn son. Lew had assumed for years that the couple were his mother's age. They were about 35, and he had roughly 50 mutual friends with them on Instagram.
He started going over on Sundays, which is the day appraisers write.
The question that started the company
And he goes, why you working on a Sunday? Why can't you have AI do it? And those were the words that essentially changed my life.
Jake Lew
He said he had always suspected there was a better way. His description of the job is physical rather than clerical.
The work that made him look for software
there's long day of inspections, coming back, typing reports, fatigued, and here in Michigan, we get snow, right? driving in snow for eight hours a day, that's exhausting.
Jake Lew
He went home, researched what existed, found nothing that met current technology standards, and started building.
2. Building For The New Form
Jeff Morley's question was why Lew went into the software business at all rather than building a private tool for his own office. The answer is that he started somewhere else entirely.
The first product was built for the legacy 2.6 form. Lew and his co-founder Jordan spent eight or nine months on a proof of concept applying AI to the 1004 and the 1073, the single-family and condominium reports.
The prototype cut data entry to minutes
we were doing those reports and data entry in like 7 to 12 minutes using AI
Jake Lew
They gave that platform away. Lew said more than 100 people were using it at one point, and the value of it was the feedback and the chance to learn the XML and work out which partners were worth having.
The pivot came at Val Expo in 2024, and the reasoning was competitive rather than technical.
Everyone starts over on the new form
With 36, we figured what an amazing opportunity here. Everyone's got to start over. Everyone's got to start from the same start line and then whoever gets to the finish line first will ultimately prevail.
Jake Lew
Against the legacy providers on the old form, he said, the mountain was the installed base rather than the product. He added that even now, with a genuinely new workflow to learn, moving appraisers onto a new platform is hard — and that it would have been close to impossible if the old form were still an option.
Verrett's own history sharpened the point. He built a report writer ten years ago and could not break into what he described as a monopoly or an oligopoly, and he now reads his own failure as a timing problem rather than a product one.
Morley's framing was that the vendors have split into two camps: some rebuilding from the ground up, others trying to make the new report feel like the old one. Lew's answer on that was aesthetic as much as strategic — the existing products looked like they were from the 1990s, and he wanted something a person could look at all day.
3. Where AI Stops
Morley asked where the red lines are on what AI should and should not do inside the software.
The claim the company is built on
Definitely for us, we believe an appraiser powered by AI will take the profession to the next level.
Jake Lew
The guardrail is the value conclusion. Lew said the platform does not touch that section: the appraiser supplies the data, the software analyzes it and returns an output, and the appraiser keeps full control. What AI is for is the copying and pasting of legal descriptions and owner names, and rewriting text into clean prose once the appraiser has given it context.
Comparable selection is the part he says a model cannot do.
Picking comps is judgment, not data work
I still believe it's an art to select comps.
Jake Lew
Research that happens off the screen
AI is great at analyzing data, but for us to be able to pick competing properties and why things sold a certain way, AI can't do that.
Jake Lew
He made the same argument about his own engineers: a developer using AI moves from typing code to reviewing it, and the human still has to catch what the model gets wrong. His team is 15 people, which he described as small against the legacy vendors and fast because of it.
His formulation of the division of labor
We always say AI is the car, the human's the driver at the end of the day.
Jake Lew
And why he thinks some jobs survive
It doesn't replace the human expertise in really anything.
Jake Lew
Asked how an appraiser should think about not getting left behind, Lew's answer was to separate the jobs. Appraisal software is his company's problem to solve; everything else is the appraiser's own. He said he has met appraisers who built their own customer systems with a chatbot.
Appraisers are already building their own tools
I met a bunch of appraisers who built their own CRM, using Claude, right?
Jake Lew
The show's counter-argument was that the risk-aversion of the industry's stakeholders is itself a protection: nobody wants to look back in ten years and find that AI was allowed to make final valuation decisions too early.
4. Paying For Frontier Models
The question was how a small company stays on the newest models when the newest models are the most expensive. Lew said there is no balance to strike, because the cost is a pricing problem rather than a product one.
The pricing model he wanted was rejected by the market.
The industry would not take per-report pricing
We don't want the best way to do it, honestly, is a per report model. That's that's literally the best way to do it. But for us in this industry, a pro report model is has never been a thing.
Jake Lew
After what he described as major pushback, the company went back to predicting how many reports an appraiser writes in a month and baking that into an annual price.
He expects the input cost to fall.
More model suppliers means lower prices
As more models come out hopefully the price will drop and we're kind of seeing that now.
Jake Lew
Being web-based is what lets him pass the newest model through: an appraiser opening the platform gets whatever is current that day, with no download and no version to upgrade.
5. A Closed Environment
The show's framing was the one the large cloud companies use — the value of an enterprise platform is that it handles the security questions for you.
Lew's version of that is a retention policy.
Nothing is kept and nothing is trained on
Yeah, we're zero data retention. So like we don't feed any we hold the data for 30 days to monitor any abuse once after that 30 days is gone that data is gone and deleted. Nothing gets sold off or trained.
Jake Lew
He contrasted that with an appraiser pasting a report into a consumer chatbot, where the data goes somewhere and is trained on. Some lenders have pushed back on the word AI itself, he said, while other employers are enthusiastic about AI and using it in exactly the open way the lenders are worried about.
The distinction he expects the industry to learn
The closed environments are the best route to go for, especially in our in our business for sure.
Jake Lew
6. Ten Vendors, Then Fewer
By the show's count there are ten approved report-generating platforms today and about another 15 on the way, against roughly four vendors for forty years.
Lew said he and his co-founder asked about the number at the conferences themselves, and that the same thing happened on the last format change — a crowd of new entrants, most of whom fell away.
Two things explain the crowd: AI makes building cheaper, and the specification is public, so anyone can read it and attempt an implementation. His warning is that passing the test is not the business.
Verification is not evidence of a working product
I am a firm believer in don't just trust verification.
Jake Lew
Verification is 15 or 16 test cases, which he called great as far as it goes. The separation, in his account, happens in live production, where cases come up that no test case covers — and in support, which he treats as a business line rather than an overhead.
He described providing that education to thousands of appraisers every day as a whole business in itself, and one that a small team cannot staff.
His forecast for the vendor count
I think next year will be a different story. I think there won't be as many left.
Jake Lew
The constraint he named is team size: a small team can pass verification and cannot carry the support load behind it.
7. Volume Is Doubling
Verrett asked what share of the 3.6 reports delivered so far have come through Aivre. Lew has the number and cannot publish it. What he would say is that there are two leaders, his platform and one legacy provider, and that there is no close third.
On throughput, he said there is no average because the line is still bending.
Volume turned in the last two months
We've really seen a kick the past two months.
Jake Lew
The monthly figure, as he put it, is doubling almost every single month, and the month in progress was going to pass the previous one within a day. The demand is coming from lenders switching over.
Two large lenders and a Rocket pilot
Two big lenders coming on board. I know Rocket's in their pilot now.
Jake Lew
He also said an email went out from one large lender saying October is when it moves everything to 3.6.
His view of the timing is that the industry left it too late.
The advice has been the same all year
our messaging has been start as quickly as possible and the appraisers that started early, they're going to be ready for this.
Jake Lew
Verrett added that he has spoken to several large appraisal firms now seeing a hundred or more of these orders a day where it was silent for months, and that hundreds if not thousands are running through the system daily.
8. Where Appraisers Get Stuck
Asked what actually trips people up, Lew gave a list rather than a theme.
Unit interior, level and room detail. The new report wants rooms broken out by level, and appraisers are bunching every room and every photo under one heading
Bedroom and bathroom counts. Below-grade rooms used to be separate; now everything is counted together
The sales comparison grid, which he called a new beast of its own — appraisers do not know to add the relevant rows, then cannot find the air-conditioning line because it is not there until they add it
Accessory dwelling units, where the subject has one and the comparables do not, and adjusting for total units inside the unit section
Amenities. The subject has a patio and none of the comparables carry an adjustment row for it
The rest of the report, he said, is broadly unchanged: the site section, condition, quality and property amenities.
What he finds is that the fix sticks.
One explanation is enough
we don't get the same question asked twice essentially
Jake Lew
The reason those rows are missing by default is that the enterprises want the appraiser to decide what is relevant in that market. Lew's read is that once an appraiser understands the row is theirs to add, the point clicks.
He expects the learning curve to end quickly once the work is consistent. Right now an appraiser might write one 3.6 report, then a run of legacy ones, then another — which is the worst possible way to learn anything.
9. Two Pages Become Sixteen
The show's own illustration of the change came from a class one of the hosts gave to a local board of realtors. He showed the first two pages of the legacy single-family form, with each section densely packed, and then broke the same content out into the new format.
It ran to 15 or 16 pages.
The reading was that the new report is far more legible for the person receiving it, and a large adjustment for the person writing it, whose whole habit has been to pack information as densely as possible and move on.
There is a physical version of the same problem. The show described an appraiser who bought a 32-inch monitor and turned it vertical so the whole adjustment grid fits on one screen instead of scrolling.
Morley's point was that appraisers are learning two things at once: where the data goes in the new format, and how their particular software wants them to get it there. The second varies from vendor to vendor, whether that is adding a row, making an adjustment or setting up automatic adjustments, so an appraiser switching platforms is learning both problems simultaneously. His advice was to take the classes and the webinars and practice writing one, even in software you do not end up buying.
10. The Emotional Response
Lew was asked about the appraisers who say publicly that they have one foot out of the profession and will retire rather than learn this.
His read on the objection
I think it's an emotional response.
Jake Lew
And what is actually underneath it
I think if you take out the emotion and you just kind of look at the report, it's a lot of the same stuff that we've been doing for the past 15 20 years.
Jake Lew
Same data points, he said, different output and a different look. The objection he hears most often is about technology, and he does not accept the framing.
The dinosaur argument
I get that comment all the time. Oh, I'm a dinosaur. I don't know tech.
Jake Lew
His answer is that the new report is a form, not a technology, and that the way in is volume control rather than commitment.
A dosage, not a decision
You don't have to take 30 a week. Take two. Take three.
Jake Lew
The same applies to the field work. His standing recommendation is to keep doing the inspection the old way, clipboard and pen, and learn the data set first, then move to mobile capture later if you want to. He described an appraiser who has completed more than 40 of these reports still working from paper, and another who was uploading photographs from an SD card.
Learn the content before changing the method
Master the data set.
Jake Lew
And the platform is not limited to the young
Then we have the ones who are 70 80 years old that we have on our platform.
Jake Lew
11. What It Does To Fees
This is the section with two parties and no agreement. Lenders do not want to pay for an appraiser to learn something they were going to have to learn anyway. Appraisers point out that there is more data in the report and that they remain accountable for all of it even when software puts it in.
Morley put the appraiser's side directly.
More data, same accountability
you've got appraisers on the other hand that are saying there's a lot more data that goes in here and even if I have a tech lift, I'm still accountable for all of the information that's got to go in the report
Jeff Morley
The show's estimate is that without an AI lift the new report reasonably takes 15 to 20% longer than the old one.
Lew's answer on the steady state was narrower than the question. Give an appraiser a year or two in the software, he said, against twenty years of practice on the old form.
At steady state he expects parity
Comparing single family to single family, I think the time will be very similar.
Jake Lew
Two-to-four unit properties, accessory dwelling units and rent schedules are genuinely more work, and he did not claim otherwise.
On what to charge, he declined to give a number and gave a limit instead.
The complaint about pricing is a complaint about neighbors
We've heard appraisers, I'm charging $2,000. It's like I don't think you're gonna get an order.
Jake Lew
The reason he gave is that the fee is paid by the borrower rather than the bank or the appraisal management company, so an unreasonable fee lands on someone in the appraiser's own market.
Morley's view is that the pricing settles itself.
Capacity, not negotiation, sets the price
I think the market's going to work itself out.
Jeff Morley
The mechanism he described is appraisers running out of capacity — two weeks out on turn times, at which point they can afford to charge more. He also expects the software to keep improving under them, and borrowed a line from the early days of large language models to say so.
Today's software is the worst version you will use
early on in the LLM days I heard the comment frequently that this is the least intelligent model you're ever going to use
Jeff Morley
Lew's version of the same point is about who benefits.
The advantage goes to whoever is willing to learn
I think the ones who are willing to learn are going to benefit the most.
Jake Lew
And the method is repetition
You practice by repetition just keep doing it over and over again
Jake Lew
The practical problem he named is that appraisers are busy and will not rehearse a report without a live order in front of them. His recommendation is to practice before the first one lands, then take as many as possible to build the habit.
12. After November 2
Verrett's rapid-fire round opened with the question of how many legacy 2.6 orders exist in 2027 and beyond. Lew's answer was none for the enterprises, and a short tail elsewhere.
The Federal Housing Administration is aligned with Fannie Mae and Freddie Mac on November, which matters for a specific reason: a borrower who cannot qualify for a conventional loan switches to an FHA loan, and the appraiser has to be able to take that work.
Where he expects FHA to land
I think FHA will be good by November, end of the year.
Jake Lew
On the Department of Veterans Affairs he said he has a date in mind and cannot say it. He has heard that something is coming.
Non-agency lending is the tail. His prediction is that everything transfers by May 2027, on the reasoning that the enterprises are allowing revisions up to that point, which gives non-qualified-mortgage lenders a grace period from November through May.
Aivre does not support the legacy form at all, which was a deliberate choice and is the single biggest complaint the company gets. Lew said the idea of building it is still discussed internally and the constraint is focus rather than capability.
Asked whether the deadline itself might move, he was flat.
The date is fixed
No, that is a hard set date.
Jake Lew
He talks to people at the two enterprises most weeks and said their test cases and numbers are in, so what is left is getting the large lenders on board — and two more arriving in the past week made the date harder still.
13. Hybrids And Inspections
The last of the rapid-fire questions was what happens to the appraiser who does the inspection, given the hybrid and desktop products coming.
Lew expects the inspection to stay with the appraiser for the foreseeable future, and expects the scope of work around it to change.
The new report was designed for split assignments
I think hybrids will become a thing and this report is built for that
Jake Lew
Verrett said the show's own third episode in this series, released the same week, is about that hybrid momentum.
Morley added the part that ought to concern appraisers more than the form does. Many loans qualify for something other than a full appraisal, and the full appraisal is often just the default. If the bottleneck becomes the appraiser writing a new report, waivers and hybrids ramp up instead — and an appraiser who refuses the new form may find the lender skips the appraisal altogether rather than ordering the old one.
14. Who Blinks First
Verrett raised a rival podcast's observation that the industry may not be fully ready, and asked whether that changes anything about the date.
Lew's answer was that unreadiness produces latecomers, not a delay, and that the enforcement mechanism is financial rather than regulatory: the enterprises stop buying the loans, and the lender is left holding them.
The motivator is the balance sheet
But when it hurts that balance sheet, they're going to be the first ones to jump.
Jake Lew
He said he has heard theories about how people will try to route around it — submitting a PDF through the delivery portal rather than the XML, or emailing a legacy report as a PDF. None of that, he said, is going to work.
Verrett's closing observation was that the appraisal profession's default posture is distrust, and that the surprise of the past year has been how many sceptics have publicly endorsed what Lew's team built. His recommendation to appraisers was to try every approved platform rather than any one of them.
Lew agreed with the advice against his own interest.
His own pitch is to shop around
Try them all. Whatever fits your workflow better and my software might not fit your workflow of your volume.
Jake Lew
Bonus Insights
Twenty years without a change is the real story
Both sides of the conversation kept returning to the fact that the residential appraisal report did not change for two decades, and one of them made the point that in any other part of real estate, or in any sport where the rules move constantly, that would be remarkable rather than normal.
The brand was built on posting
Verrett told Lew he is a household name among appraisers. Lew credited a head of marketing named Nathaniel and a decision at the start of 2025 to build the brand by going where appraisers already are: Facebook, LinkedIn and YouTube, with a weekly article. He said it took about 18 months to work, and that the reason it worked is that nobody else was doing it.
The product exists because a designer made it look modern
Lew was repeatedly clear that he was the visionary and his co-founder Jordan was the one who made it real, and that the fresh, modern look is the reason appraisers give it a second look at all.
Listening as the moat
Asked what the company does differently, Lew's answer was feedback: more than 100 appraisers used the free legacy-form product, everything they said went into the new one, and he describes every appraiser on the platform as having had a hand in building it.
Lew's bottom line is that the November 2 date holds, the appraisers who treat the new report as a form rather than a technology will absorb it inside a year, and the crowd of new software vendors thins out as soon as passing a verification test stops being the same thing as running production volume.
Products, Companies & Tools Mentioned
Aivre (Lew's company — the first appraisal software the enterprises verified for UAD 3.6, a team of 15, web-based, zero data retention)
Rocket Companies (Named as being in pilot on the new report, one of the lender moves Lew says is driving his volume)
Fannie Mae and Freddie Mac (The two enterprises behind UAD 3.6, the verification process and the November 2 date Lew says is fixed)
Federal Housing Administration (Lew expects it aligned with the enterprises by November, which matters because conventional borrowers fall back to FHA loans)
U.S. Department of Veterans Affairs (He says he has a date in mind for its adoption and cannot give it)
Claude (Named as what appraisers are using to build their own customer-tracking systems outside the appraisal workflow)
Delta Air Lines (Where Lew's mother worked for 28 years before the buyout that led to the nannying job behind the company's origin story)
Michigan State University (Lew's degree, 2013, before he qualified as an appraiser)
Books & Resources Mentioned
Uniform Appraisal Dataset and Forms Redesign (The public specification for UAD 3.6 — Lew's point that anyone can read it is why so many new vendors appeared)
The Appraisal Update (The podcast Verrett cites for the view that the industry may not be fully ready for the deadline)
Listen to the full episode
🔴 YouTube | 🔗 Episode page
Watch the full episode:
If this was worth your time, send it to someone closer to the industry than you are.
Get the latest market chatter as it happens: