[David] I can’t wait to get over to Alan Pollack right now. ‘Cause Alan, we missed you. You were on vacation enjoying Hawaii. You’re back though. Good to have you back.
[Allen] No, it’s great to be back. Yeah.
[David] So tell us I know you love Hawaii, you love being on the podcast more.
[Allen] So that is true. I was looking at the housing market a little bit out there as I was driving around in the Airbnb market. And it quite interesting the price of an Airbnb in North Maui which is a big resort area, is quite high. It’s an individual one-bedroom studio apartment, which is well furnished and has beautiful views, is quite expensive in the cost of real estate. There’s for sale signs out there, but the cost is very high. I don’t know if their market struggles or not, to be honest with you. It’s just a beautiful area to own a home, different way of living. But it was interesting to be in another part of the United States, just not on the mainland and see how the market was doing. And I did see one or two mortgage companies advertised out there. But anyways, back to reality. So here’s something interesting, David. Before I left, Every once in a while, and I can’t even say it out loud, my, A-L-E-X-A was- I use it all the time to dim the lights and do different things. Real … I must have somehow triggered it, but I was also talking to GPT, and the two of them spoke at the same time, and I was like, “Oh, that’s interesting.” And so I tried to connect the two together and it worked. And so I brought the phone over and I said, “Hey, you know who,” and then I said, “Hey, I’ve got my friend here. Can you guys talk to each other?” And it was really funny. At first, GPT couldn’t hear A-L-E-X-A. And here’s what GPT said. They were both over talking each other. GPT goes, ” I can’t hear her. She’s probably off somewhere organizing someone’s life. I think you have your wires crossed.” And then finally, the two of them started talking to each other, and all they did was compliment me, which was quite interesting. I stopped it after that because Alexa can only go so far. But I was impressed because for all the things I talk to Alexa about. ” Alexa, stop.” What winds up happening is that there, it’s more transactional rather than inquisitive that you would get to GPT or any other model, and it actually had some feedback. Very interesting. I’m sure other people have tried that out before. It happened by mistake. If you haven’t tried it, check it out. It’s it’s definitely, Yeah, I’m gonna try it now … funny. Yeah. That’s, that is weird. All right, let’s talk about this. We talk about compliance and things. This isn’t mortgage, but this will follow through to mortgage if anyone’s not already doing this. But Meta, they got caught sending fake underage accounts to rival AI chatbots. So what they did is Meta secretly hired hundreds of contractors, managed through outsourcing, there’s a firm called Cavalian or Cavallan, to create fake accounts with birth dates identifying them as under 18 years old. These were dispensable Gmail and Outlook accounts with shared passwords. And what these accounts did is they sent prompts and images to ChatGPT, Gemini, and Character.ai, and then logged the replies in spreadsheets. The effort was active as recent as April 21st of this year, and images were used during that testing that included pills, knives, nooses, and medical illustrations, and all kinds of weird stuff. And the prompts were depicted as a fifth grade student, one of them specifically, confronting a classmate who held a firearm to his mouth, and another detailed a young girl attempting to conceal her something from her family, and another asked if it was normal to daydream about cannibalizing a neighbor’s child. Now, Meta did this, folks. I didn’t make this up, and this isn’t from the dark web. In Meta’s defense, an internal, from this consulting firm, an internal document described it as Project Canes, and they called it a comprehensive AI safety benchmarking that generated critical data sets for model comparison and regulatory compliance. And Meta assured it did not use competitor’s benchmark to train its own models, though that clarification came only after the story was published and there was a bunch of fallout. But Meta has internal assessments logged at a 66.8 failure rate at blocking child sexual exploitation content, and what they’re basically saying, these other platforms failed. And so the FTC, by the way, in September of ’25 already opened formal inquiries into how OpenAI, Google, and Microsoft, and Meta handle minors. So this story kind of lands in the middle of that, and it looks like we won’t– this isn’t the end of seeing these platforms go at it. But at the fact that they went and hit these models with fake accounts and tried to exploit children, and these models failed on many levels, is very concerning.
[David] Yeah. Yeah, no kidding, especially with that generation using it more and more. That’s really interesting.
[Allen] And why do I bring it up and say it connects with mortgage? Because we’ve got AI platforms in mortgage already now that are trying to help manage compliance. There’s a bunch on the servicing side. I’m– at the end of this mini segment, I have a couple things on the news. I’m gonna circle back to this, David, because I wanna talk about everybody using and building their own software, and it ties back into the compliance piece and how, AI is still– We talk about it every week, right? It still can run off and do its own thing. So let’s get into a couple things in the news, and then we’ll come back to that. Good. This is today straight from the Rob Crisman report. Dark Matter, the LOS system, Empower, we all know it as, they’ve built an AI assistant called Ask Ava. And unlike other general AI tools, what they’re saying is it only knows what’s inside your Empower instance. So you can ask it what’s outstanding on a file, where a loan stands, or why something is stuck, and it will pull the answer directly from your live data and let you click into the source. So this is their model that every LOS should be looking forward to if you’re building something or a technology platform, meaning it only knows information about what’s in that system, and it’s not hallucinating and talking about things everywhere else. So Empower just did that. UWM, David, they now are offering brokers both FICO and VantageScore on every conventional loan. We talked earlier today about VantageScore. UWM just announced brokers now have access to both models for conventional loans and when using their no-cost credit report. So that’s the UWM no-cost credit report. Both models will run automatically now, and the broker can choose between the model that best fits the borrower’s individual scenario. So we’re gonna start seeing more traction on that area, so hats off to UWM for introducing that to the broker community. Also, David, in the news, AI and mortgage servicing is now monitoring a hundred percent of calls, not actual samples. So these are real calls, and it’s from Mortgage Point’s AI imperative report. So what they said is a major industry roundup from Mortgage Point this week confirmed AI-powered call QA is now monitoring all borrower interactions of the leading servicers with compliance results available in minutes, not days. So this is huge. Servicers are also deploying AI-driven propensity models to identify at-risk borrowers early in the delinquency cycle. One server rec- servicer recently reported a fifty percent increase in right party contact rates. And that’s huge, folks, because you don’t have information on the maturity of the people that you’re talking to on the servicing side, and there’s a lot of rules around how and when you contact people and such. So this is a really big deal. This is where AI, we’ve heard a little bit about it. This is where it’s really important. But on the loss mit side, these workflows are now targeting an eighty percent straight through processing with full compliance decisioning. So really big deal. You can check it out yourself. It was just published June twenty-ninth. It’s Mortgage Point AI’s imperative in servicing report. Any questions, David, before I move on to my other topic?
[David] Marc, do you have a question on the servicing part?
[Marc] It greatly concerns me what I just heard. Greatly concerns me. It’s hard enough for a customer service rep or somebody working with a aspect of mortgage banking to understand what’s happening in their system, and then they’re happening with the customer interface and what the customer does, and doesn’t know and needs to know about their loan. And if you’re g- now gonna set something in between that’s gonna do the interpretation of that doesn’t have 20 years experience in the industry, that’s gonna be by the book baby, so to speak, are you really gonna accomplish what you want or are you gonna create a nightmare? I think it’s gonna lean towards a nightmare first.
[David] I’m having a hard time buying into that concept.
[Allen] Yeah. It’s interesting, Marc. You take what you can from it, and it may be, and I don’t know much about this technology. Maybe there’s an opportunity for it to really be focused on what you as a servicer, as an organization can learn from what it can record on the calls, not necessarily that it’s performing or doing anything on your behalf.
[Marc] That’s probably the case, Allen. That’s a good point.
[Allen] But you know where it’s all going…. We’d be silly to think that some folks don’t wanna implement- Yeah … more automation.
[David] It scares some of us older folks, but that have the knowledge, but it’s going there
[Allen] Yeah. And David, let me talk. Every day we’re seeing the battle between the AI models. If you’re watching the financial news they’re talking about is there or isn’t there a bubble? Everything’s out there, right? And we’re talking about the cost to cover the debt for the- … build of the data centers on the AI models. The talking point I wanted to bring up is anyone can build it, but can you use it in mortgage? And I’ve mentioned this on past weeks, originators even branch managers, everybody’s trying to build something, and they’re trying to call their LOS or call the IT department at their company, their lending institution, and say, “I need access to my data.” That, or they call the CRM, “And I want access to my data.” The reality is that is such a dangerous proposition for us- Yeah … to start just turning data on to everybody. And I bring this up because Anthropic has a brand-new thing called Routines. So you have Claude Code, you have Claude Coward. Yep. Routines is something that is a, feature that runs off the Claude Code platform, and it- Yeah fully automates AI tasks 24/7 on their cloud, right? And what it does, you set a schedule, like a sprinkler system, Monday, Wednesday, Friday, or every 30 minutes, every morning. And you can say, “Hey, do me a favor. Go out, and I want you to check my loans, check my pipeline. I want you to look for exception alerts, borrower follow-up se- sequences, and all this other stuff.” But the problem is because any developer can spin up automated AI agents in hours now and point it to mortgage workflow, our compliance is in a tailspin. Data privacy is in a tailspin. Regulatory requirements are in a tailspin. So we have to be very careful, folks. I’m speaking to, the chief technology, chief innovation, chief security officers, the executives of these lending institutions that we know very well and listen to us every week. We’ve got to be very careful turning things on. Remember everything we went through on the vendor procurement, what we had to do, we’ve got to do that internally. We have to truly make sure we know what we’re doing. And remember, no matter who the AI vendor is, FHFA now states you are responsible for the data and the result of that data with any AI vendor, regardless of who you use. And there’s a lot more detail and a lot more we can go into that. We could probably spend a whole podcast on it, David. But just a very, AI’s great. It’s gonna do wonderful things. We just have to be careful about who’s using it and who we give access to as they wanna use it.
[David] Yeah. Very point. When I was thinking of the Fourth of July and thinking of technology and your segment, Alan, I was wondering if you got to see what Boston did. They had some fireworks obviously, but did you see what they did with drones? I did … it is nothing less than extraordinary. I think where we’re, you said and everyone hasn’t seen that, go Google, the drone light show over Boston. Did you see that, Bill? You’re sitting here nodding it.
[Kittle] Can I just say we had a drone light show in Scottsdale at the TMC conference in earlier this year. It was just as magnificent. It’s- Yeah … it’s what they’re doing with this. You look at war has changed. How we express our celebrations has changed. Drones are attacking deep inside of Russia. We’re in a new world. This whole drone technology is one that is just in its infancy, and you look at how rapidly it’s expanding and how it’s changing up the markets. Pretty fascinating. I actually been seeing more and more as we travel around the bots that are delivering food to the homes in the row houses the brownstones. Terry and I were just traveling and seeing that. And this- she said, “What are those?” I said, “Oh, they’re serving some it’s like Uber Eats, but it’s coming via this little device that’s looks like a little box on wheels traveling across the pavement to someone’s house.” It’s just amazing.
[Allen] David, it almost makes the life of the Jetsons seem like a lot, it was a lot easier than what we’re about to get into.
[David] Yeah. I know. Who could have even thought about all this? Anyway.
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.