AI, Data Risk, and “Vibe Coding”: The Mortgage Industry’s Next Compliance Challenge – 03/31/2026 Weekly Mortgage Update segment

AI, Data Risk, and “Vibe Coding”: The Mortgage Industry’s Next Compliance Challenge – 03/31/2026 Weekly Mortgage Update segment

[Alice]  All right, next up we have Allen Pollack, COO, consultant and tech expert. And Allen, what have you got today for our weekly tech report?

[Allen] Oh, thanks Alice. Just a little bit of this and a little bit of that. I’m gonna start with something funny that so after we got off of the David Kittle Chat GPT birthday joke this is the last thing it left me with, and I didn’t wanna take focus away from his birthday, so I saved it for this week. So here it is. Why did Mark Zuckerberg buy Instagram for a billion dollars when he could have just downloaded it from the app store for free? So it’s one of those, good shower thoughts. Yes. All right. Shower thought. Yes. Here’s another great thing going on with ai, in the world of AI taking over and doing everything it’s taking over the world that some people think seriously. Google AI told people to put glue on pizza. So Google rolled out its AI search engine answers recently. Now, when I say recently, it could have been about a month ago one response suggested people should add glue to pizza sauce so the cheese would stick better. It happened because the AI scraped a joke comment that it found on Reddit, and it treated it like real cooking advice. Wow. Marc and I always laugh at how AI can be wrong. And here’s an exact case where you cannot just take exactly what it is that you get from AI and run with it. Another thing that, just great advice, there’s a guy, he and I talked about this on a prior podcast. He said to ChatGPT, do me a favor. What I’d like you to do is count to 100 without stopping. And it went 1, 2, 3, 4. Okay, I’m gonna tell you number five, but then I’m gonna keep going to a hundred. And he is no, do it again. Don’t stop and he couldn’t get it to go to a hundred without stopping. Then he went to, it was either Gemini or Claude, and he did the same thing. And Gemini couldn’t do it either. And then he put the two of them to talk to each other. And his point was is they don’t always listen. They do what they want. It’s prompting and how to talk to AI is critical. So I know a lot of organizations are rolling out private clouds of open AI and other technology tools that have AI baked in, including Microsoft copilot and such. You gotta really make sure you’re training your staff and teaching people that what they put in is not always exactly what they expect to get out. So keep an eye on that. A big thing going on right now talking about vendors and data. And what’s going on in the mortgage industry is called Vibe Coding. It, the virality of that word is everywhere. AG Agentic experiences is everywhere. Basically you chat on the left side of the screen, you say, Hey, I’d like you to reach out to these borrowers for me and tell ’em it’s Taco Tuesday. Send them some coupons, and it automatically would go and find the borrowers based on the data set it has access to. It would automatically take the credit card, go buy some Taco Tuesday discounts at the local Mexican place, wherever it is that they live automatically, and send out those emails and it have tracking code and links and all that in it. The thing is, everybody’s trying to get access to the data. Everyone’s trying to do things like this Vibe coding is this world where it’s part agentic and it’s part. Any of these platforms that are out there are giving you the ability to code your own systems. And so there was an article out called, is It The Death of SaaS? That’s software as a service folks. And what they’re saying is, any, why would you pay for a very expensive CRM as an example when you can go code your own in a matter of two days or one day you have access to the data. So we have a lot of challenges on our hands in the mortgage industry. We have to deal with the fact that anyone that can get access to data, like a loan officer for example, what system are they putting the data in? Even if you don’t know, I’m speaking to the executives at each different organization, depository and non-depository. If there’s an Excel spreadsheet that your originators or your staff has access to and has borrower information on it and they upload it to this new vibe coded platform, they have to reach out to borrowers. You’re in significant risk of major lawsuits and data breaches. If just one spreadsheet with two borrower names are on it, there’s a full investigation that has to happen. You have to call out to your insurance policy, you’re going to need forensic account forensic data scientist, by the way, they’re 500 bucks an hour, gets very expensive. So it’s time to really be careful about vibe coding and agentic coding and these experiences that your staff has access to, to make sure that you’ve got strong policies in place about the usage of your systems and your data, and that you’re not just randomly getting data out there. I can tell you that. A lot of the, so it’s kind, it’s yeah it’s a seesaw in a way, right? You’ve got a problem on the left side and you’ve got a solution on the right, but if you take the solution on the right, you have another problem on the left side and more specifically what I’m referring to, Alice.  Is that we have all these people that can go build these systems where they come with their own CRMs and they want access to data, but sometimes the data’s so difficult to get that they’re sneaking around a compliant way to get the data to go do what they want. So you may not even know they have Excel spreadsheets of the data or they can’t, you brought in LO from another company and they brought an Excel spreadsheet off of their personal thumb drive, which by the way has been in their backpack in the back of their car for three weeks and it’s got a thousand people on it. There’s no way to know. So if there’s a better way to control access to data. Now I’m speaking to the vendors. The vendors in our industry are doing a very, mark, this is one of my rants. They’re doing a very poor job of making data accessible. The vendors want to charge a ton of money for anyone to get access to the data. They wanna make it almost impossible to put a service in the industry because everybody wants a piece of everyone’s money, which started, by the way, in the two thousands when Ellie Mae launched epass. They were one of the very first ones. They went out there and said, anybody that wants to transact through our system, it’s 15 or 30%. Of your invoice, and I was the very first pricing engine vendor on epass in 2002. And no matter how many people, one person just clicked to see what the Ellie MA interface was, I had to give up. It was, I think it was 15 or 20% of the entire customer invoice. Things have changed. That was an old company, by the way. They don’t exist anymore. I’m just using it as an example. This industry has to change. We cannot be overcharging people for access to their own data. We can’t put too many complexity or controls in place. And then we have some vendors that won’t even provide access to the data because in technical terms, they’re not ready to do it. Folks, that’s a huge red flag because now you’ve got a vendor that has your data and you didn’t even know they don’t have the right systems to give you the data. And are they open to a data breach or what are they trying to do or build? There’s a lot here, so I don’t mean to mention that to scare everybody, but it is my rant. I’ve never had a rant before, but we gotta do a better job as an industry to make the data available so we can put better compliance and controls in place and we have more access to the data and we can do more stuff. So that’s really all I have for today, Alice. Oh, I will go down to one comment that was made earlier. We don’t do a good job in communicating. It’s funny that it’s in the same direction of where I’m going. We can’t do a good job if we don’t have the tools we need to do a good job. You could only do so much as a person. Now, as an lo you should be making personal phone calls to people that you closed. But seven years out, it’s a little bit harder, right? And it’s a lot of phone calls. So we need the technology to help us, whether it’s telling us who we need to reach out to or re-engagement campaigns, but we need better access and better tools. So we need to do a better job all around.

 

[Alice] And Allen, that was fabulous. I have to tell you, I’m sure when you were talking about the data and the challenge with getting access to it from vendors, there are a lot of people listening to this right now who are shocked that it’s not their data. In some contractual arrangements or that there’s any issue at all with getting your data. And I have seen this at different companies over the years where you end up realizing, wow, they think it’s their data and it, so you have this disagreement in the first place. And then when you, if you can come to terms that it is your data as you described in detail it can be scary with trying to get it back. So for those of you out there who are realizing this for the first time it is an education for many of your frontline folks who don’t realize this, they always think it’s mine. Loan officers always think the borrower data is theirs. That’s an education a lot of time that it belongs to the company. And that making sure that relationship between the company and the vendor is clear. So if you don’t have that ironed out, you need Alan Pollock, right?

[Allen] Yes. Thank you, Alice. Very well said. That is very true what you said.

[Alice] Thank you Allen, so much for all the enlightenment that you just gave to us about the markets today. It’s been great to have you here and all the insights. Anybody else have anything they wanna add on to Allen’s segment? Any questions for him?

[Bill] So Alice, I think the only thing that I would add is I still talk to people and balance’s point about uploading spreadsheets and folks take, still have the opinion. If I take one piece of data out of the spreadsheet, then I’m clean. What? Not recognizing. How much other data is out there, and now how easy it is to match up different data sets to say, okay, so you think you’re safe by eliminating the borrower’s name? Guess what? They can go pull this data from here and know exactly who the borrower is. So you haven’t done the right thing. And a’s point on that, the culpability, that hasn’t gone away just because you thought you did something that was creatively smart.

[Allen] Yeah. And actually it was a great point you made. Without getting into detail everyone’s heard the term PII or personally identifiable information. What that triggers is just a very name and email is PII data. It’s a very small data set, especially in the banking world. For those of you that think like you just said, you can just remove one piece of data. Or a couple pieces of data you still can be triggering compliance with having PII in your spreadsheet.

[Alice] Thank you, Allen.


Allen Pollack, Chief Operating Officer, Tech Consultant

Allen Pollack, a Mortgage & Financial Services Technology Advisor, is a subject matter expert in the mortgage origination process along with software product management and software development.

In today’s financial services push to all things Digital, Allen has been helping lenders and financial services solution providers align their digital transformation and technology strategies by removing the human element of risk, and automating processes that drive efficiencies and margins into profits.

Over the course of his career, Allen has co-created and developed technology business models that have birthed highly successful, innovative solutions and companies.

Allen co-founded and served as CTO of New York Loan Exchange (NYLX), a loan product eligibility and pricing engine (PPE) that made an immediate impact on the industry, scaling the company quickly and forming partnerships with multiple mortgage and financial lending companies. In 2012, Allen was a co-founder of a merger between NYLX and Aklero Risk Analytics that created LoanLogics, A Mortgage Loan Quality and Performance Analytics company. Allen served as CTO where he continued to bring new and innovative product solutions to the market that made a significant impact to mortgage lenders that reduced risk, scaled business channels, and grew profits in a very competitive and highly regulated market.

Allen is also is mortgage and finance technology contributor on a weekly live industry podcast, Lykken on Lending, and is launching a new podcast soon to be released, TechStack Radio, dedicated to technology and innovation in Financial Services.