loanDepot CIO on the evolution of build versus buy over 14 years

[ad_1]

Editor in Chief Sarah Wheeler sat down with George Brady, loanDepot’s chief information officer, to discuss the company’s evolving tech strategy and how he thinks long-term about tech in a rapidly changing environment.

Sarah Wheeler: You joined loanDepot in 2021 from Capital One. What drew you into the mortgage industry?

George Brady: You could argue I joined at the absolute worst time! And Anthony [Hsieh] was very open with me, saying “We’ve had an incredible run and it can only go down from here. It’s going to be really challenging, so strap in.” But for a technology executive, coming in at a time like that is where you get to pull back and calibrate with the executive team on what’s really important.

As a 14-year-old company, there’s a lot of technology and it was a great opportunity for me to come in and look at the foundational technologies the company was built on. And then look at where we needed to make tweaks and adjustments to set the company up for the next decade and beyond.

SW: What’s loanDepot’s build versus buy strategy?

GB: The whole mortgage tech ecosystem was very different 14 years ago when the company was founded, so there was, understandably, a heavy bias towards build. The company was founded on an LOS and a dialer technology back then. A lot was built around those foundational technologies for what we have today in the Mello system, in the stack that our loan officers love and appreciate.

Coming in now, you’ve got to look at the whole ecosystem and what’s available. What do third parties offer; where’s their maturity, where’s the competition? So we have evolved from a heavy bias towards building to one that is much more balanced. And we definitely look to buy and integrate where it makes sense.

If we have something we’re going to build from scratch, we look at how long it would take us to build a competing capability and is there really enduring value? Because the question around build should be: am I building something that’s going to provide some competitive advantage for the foreseeable future? Not just for the near term, not one to three years, you’ve got to see value much, much longer than that.

And for buying we’d want to see competition in the marketplace, quite honestly. So if there’s only a single provider in the marketplace, that’s a little less appealing than if there’s multiple providers. If you’re engaging with a third-party partner and things go south over a longer period of time, you need an easy exit ramp off of that relationship and onto something else, so you want competition in the marketplace.

SW: Let’s talk about that horizon. How do you plan for five, 10 years out when tech is accelerating so fast?

 GB: You certainly can’t predict with 100% certainty what’s going to happen in five years — one to three years is hard enough, especially as you look at emerging technologies like Generative AI. A loan origination system is a big decision, it’s kind of a decade-long decision. And so when you make a decision on a foundational technology like an LOS or a customer relationship management system, that’s a core capability for a mortgage company. So you’re trying to predict where that technology will go. Who’s leading today and who might be leading tomorrow?

SW: Is 14 years a long time when it comes to mortgage tech? Does that qualify as legacy technology?

CB: Any company that’s been around for more than a few years has some legacy technology they’re dealing with — what is called technology debt. You’ve got to be very rigorous about how you handle that debt and how you handle ongoing modernization of the platforms that your business depends on.

A good example of that for us is the LOS. We’ve got a big effort underway now that will take us through the end of next year to upgrade the LOS that we have heavily customized and that’s served us incredibly well over the first 14 years of the company. I looked at continuing to customize and enhance this LOS that we have, which has as its foundations the Empower LOS, or going to a more commercial grade LOS from Dark Matter Technologies that is more cloud native and from the ground up for that. And is there a combination of that makes more sense?

And so that’s an example where, again, 14 years ago, it made all the sense in the world to customize the heck out of that LOS and now we’ve got an opportunity to modernize not just the LOS, but all that surrounds it — our point of sale system and our warehouse management system, everything that plugs into that LOS now is getting refreshed and modernized as part of that journey.

When you talk about technical debt, there’s a lot that you build up — there’s shortcuts you take over a 14-year period, there’s things that you deploy to be first to market that you wish, in hindsight, you would have done a little bit differently. We get to clean up a whole bunch of things and really strengthen the core foundation of the technology, and the data infrastructure, which really sets the company up very, very well for the next decade and beyond. And that data foundation is a precursor to success from an AI and machine learning perspective.

SW: You mentioned generative AI. What is loanDepot doing to leverage AI right now?

GB: We have a few different things that are in the pipeline and I would tie them to Vision 2025, which our CEO, Frank Martell, outlines as a kind of corporate strategy. One of the pillars is around driving growth, and it’s customer acquisition, customer retention. And then cost optimization, and efficiency. On efficiency, let me start within the technology organization, because software development takes a lot of effort. Developing quality code is in everybody’s best interest, so one of the really exciting things with generative AI is a change in the paradigm for how software gets developed.

Back in the early 2000s, there was kind of a pair programming model was the big innovation at that time. And you would take two kinds of engineers and sit them side by side. And they would basically work on a problem and an algorithm together — the thinking being that two minds are better than one. And now you’ve kind of got this digital copilot that’s coming alongside each of your engineers, that’s actually recommending improvements to the code that the engineer is producing, as well as recommending automated test cases around the code. That’s a really exciting use case and one we are very much dialed into for this next year.

There’s also some low-hanging fruit tied to customer relationship management and how we think about leads and lead management. And it could be very simple, just thinking about, what’s the next best opportunity for a loan officer? And how are we teeing up those opportunities for them, whether it’s a digitally sourced lead, or it’s a customer that they already have, presenting that in a way so that so the loan officers are making the best use of every moment.

And then we just started to roll out a gen AI chatbot for our servicing customers, which just makes search and discovery of information so much easier for our customers. A Gen AI bot is a low-friction way to engage with a customer where you can start to answer questions they have, whether it’s around credit quality or about loan types, and what it prepares them for is for that first conversation with a loan officer.

SW: Tell me about the journey of developing a fully automated underwriting engine.

GB: This is one we’ve been testing through our underwriters over the last year and it’s everything from doing full digital verification, to providing a really, really customized set of conditions for customers so we can take what has been a process that has taken days, down to minutes to know that you’ve got conditional approved for a loan. Or even if you’re not approved, you know what actions you need to take.

We will just continue to dial up the capabilities of what that engine can do over the next year. And the amount of loans that we will get pre-approved through that will just continue to increase over the course of the year as the algorithm gets better and better.

SW: What is the main innovation there?

GB: We spent a lot of time on the rules-based system that underlies the engine to make sure that it was right. But there’s also a lot of time invested in that digital verification and the scan capability and making sure that we’re getting good quality digital data on the other side of that. So it’s been a multipronged approach to get us to this outcome and we didn’t want to announce before it was proven and baked in our environment for some period of time. We took several quarters last year to make sure that our underwriters felt really good about it, that our risk and compliance folks felt really good about it. And it’s a change in process too — it’s not just the tech itself, you’ve got to get people comfortable with the change. It is a complete team effort to deliver on an outcome like that and there’s more to come.

SW: what keeps you up at night?

GB: I think if I said anything other than cybersecurity, you should probably wonder what I’m doing! The last couple of quarters have been really troubling because of the marked increase in the number of cyber incidents impacting the industry. So it’s obviously a concern for me and it should be a concern for every company in this industry.

We’ve been fortunate that we haven’t been impacted, so that’s great. But we also have a supplier network that we’re dependent on and you’re only as strong as the weakest link in your supply chain, so we are super focused on that. I think one of the things that is going to help the industry is the was a SEC mandate around cybersecurity disclosures that went into effect in December. That means that public companies now have an obligation to disclose cyber incidents, whether those are data breach incidents or system breach incidents, within a fixed number of days through a disclosure to the SEC. That is that is raising visibility in a way that I think it hasn’t been there before, which is good.

These cyber incidents shouldn’t be secret. But it also requires changes to the way a lot of companies work inside. So to meet those disclosure requirements, executives and board members have to be engaged very early on in a cyber incident process. Which means developing a good process for how you escalate those things and how you discuss them to how you disclose them. Maybe some companies haven’t really spent a lot of time thinking about it. Rest assured —everybody’s spending time on it now.

SW: What are you looking forward to in 2024?

GB: Rate decreases more than anything! And hopefully improvements in the real estate industry for this year.

For loanDepot, the big LOS migration that we announced in 2023. We’re in the throes of right now and it probably won’t go live until early next year, just given that we’re a public company. That change will simplify a whole bunch of things for us and our operational processes. It’ll just make us better, stronger, faster and ready for an upturn in the market.

Loan origination systems is the engine that runs a company so you can’t change it out when the  business is really, really good. So these difficult times are actually the best time to contemplate the foundational technology changeouts that you want to make. And as I mentioned, we’re going to do more with customer acquisition and our Mello platform.

[ad_2]

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *