Optimal Blue’s Joe Tyrrell on deciding where to use gen AI

Editor in Chief Sarah Wheeler sat down with Optimal Blue CEO Joe Tyrrell to talk about how the company decides where to deploy generative AI and why return on investment (ROI) for customers is paramount.

Before arriving at Optimal Blue this year, Tyrrell served as president at ICE Mortgage Technology and as chief operating officer at Ellie Mae. This interview has been condensed for length and clarity.

Sarah Wheeler: What differentiates your technology?

Joe Tyrrell: There are three things that guide all our tech decisions. Most important, is it improving the ROI of our customers? We have these incredible solutions, in some cases with hundreds of capabilities that the customer has access to.

If they are only using 12 or 13 of those, we are not maximizing their ROI. How do we help the lender get the highest rate of return on their dollars? We want to become more than just a vendor; we want to become a partner in their business.

Secondly, we never want to automate a bad process. We want to solve a problem, not just automate. Third, making sure that any use of gen AI makes things better, not worse. We see lenders making major investments in gen AI, but when we ask what problems they’re solving, they don’t have a clear answer.

As a former chief credit and risk officer, I understand the consequences of introducing unintended bias and how a mistake continues for the life of loan. So, even if a lender thinks an AI use case is cool, if the compliance team or operational staff has any concerns, everything stops. With gen AI, we can articulate the very particular problems we’re solving.

Wheeler: How does your background in lending — and lending tech — inform what you are doing now?

Tyrrell: I was a loan originator. I ran production, operations. I was an underwriter. I ran a lock desk. I’ve done all aspects of the roles our technology supports. I know what a lot of loan officers do — they know two or three programs really well, and they are going to use those and then stop.

But what if loan program No. 21 provides a better rate or payments? We can provide a gen AI assistant there to partner with originators, to serve up all the qualifying programs and tell them, on this program, if borrowers improve by 10 points, here’s a lower payment — on one screen.

You can’t ever build technology in a vacuum. At Optimal Blue, we will never use the product that we built, so we talk to the people who will and get them involved early in our process.

Wheeler: What else are you doing with gen AI?

Tyrrell: A lot! We’re the only true platform in the secondary market that has PPE on the front end, compliance, and then hedging and trading — no one else has all three. So, we look at all of these personas. If you look at the back end, hedging and trading, that’s how lenders get all of their revenue. They are no longer charging 2 or 3 points to consumers, so this is where the money is made — it’s made in the secondary market.

So, how does Optimal Blue help them increase their profitability? We started with a profitability assistant. Every day you have execs in capital markets look at, what did I lose or gain on sales? To do that, they might have to pull up four spreadsheets and get data out of their system. Some execs can do it in 20 minutes, for others it takes two hours. This is a perfect use case for generative AI and it’s already available.

The profitability assistant sees all the data, knows all the data and calculates the gain or loss. Then you can ask it a question about the data: how it compares to yesterday or three months ago, and it can show you instantly — and then remember what you asked for and include it the next day automatically.  We just saved you 20 minutes, or three or four hours, at the very start of your day, so you can take action and make impact.

Originators are on the other end of the spectrum, but gen AI can show them the 15 loan programs, show where they do or don’t qualify for better rates, and what it would take to get there. This is being piloted and we have more AI assistants we’ll be releasing at our February user conference.

Wheeler: What is the ultimate goal?

Tyrrell: Our goal is to give lenders the ability to say yes to sell profitable loans. There’s a ton you have to unpack to do that, and then it has to be competitively priced, which is dependent on what each company’s costs are, what the pull-through rate is.

We’re the only one that from the very first time an originator wants to price a loan, we can tell that company in real time with the borrower your LO is engaged with, what the likelihood that loan will pull through. We have all this data. It could be based on DTI, location, age, whether they are a first-time homebuyer, etc.

And if you know going into it that the pull-through of that loan is more likely than average, you could real-time price that loan differently. Instead of a standard rate sheet, if you think there’s a higher propensity to close, you might want to price a little lower so they don’t shop around. If you did that and locked them, you could start planning to sell that loan 60 days before. Most companies are operating at the point of close; we’re operating at the point of thought to capital market execution.

Our AIa tools are focused on the back end and front end, because if you get the front end and back end comfortable with AI, it’s much easier from a trust and adoption perspective. Then you get buy-in from the company and you can just be solving problems all the way through the loan.

Wheeler: What are some of the changes you’ve seen in tech over the past 18 months?

Tyrrell: This time last year, I was CEO of a company called Medallia, and we were working with big global companies taking advantage of technology for customer experience platforms. I learned so much in that year and a half, including that you have to understand who the key stakeholders were. In the mortgage industry, if you solve problems for the front end and the back end, they will become the mavens inside the corporation.

The other thing that changed is the evolution of large language models (LLMs). When gen AI first came out, some companies were scared to death to use it. Am I training public models for my competitors? Now LLMs have advanced so much and lenders have so many options.

Another thing with technology is that lenders are coming off a period where it’s been really hard, where anything they do introduces new cost. Optimal Blue is taking the opposite approach. All the gen AI, machine learning, all the automation — it’s at no incremental cost. If we truly want customers to grow ROI, we have to commit to giving more value to them at no cost.

This is one advantage we have because we’re backed by a company that doesn’t do earnings calls. This allows us to invest and deliver value and we’re not worried about a monetization strategy. Our monetization strategy is retention, delivering value for our customers.