TidalWave CEO Diane Yu on building an AI-first company

Editor-in-chief Sarah Wheeler spoke with Diane Yu, co-founder and CEO of Tidal waveto talk about the benefits of building an AI-first company in today’s business environment.

Yu founded and sold the ad tech platform Freewheel Unpleasant Comcast in 2014, and she served as Chief Technology Officer at Better.com from 2021-2022.

This interview has been edited for length and clarity.

Sarah Wheeler: What’s the benefit of building an AI-first business from the ground up?

Diane Yu: It is very fortunate that we started TidalWave at the right time: when the technology model for digitalization in the industry was just starting to develop and at the same time generative AI was really coming to the surface. As a technologist and engineer, I have been following this trend for a long time and then realized that this could be the very last piece of the puzzle.

Now we can build something unique from scratch – an AI-powered co-pilot engine – without the burden others have of outdated software to adapt to. We were able to move so quickly and so quickly that when we came out of stealth, we were able to get approval for integration Fannie Mae And Freddie Mac.

SW: What sets your technology apart?

DY: In order to apply generative AI in the mortgage sector, it must be 100% accurate and non-hallucinatory. And this is why we consciously chose to build an AI engine, without using any kind of ChatGPT wrapper. A lot of companies say they have something similar, but they were building a package, and that wouldn’t work because there’s no way you can stop the hallucination problem. So that’s No. 1. We chose a more difficult path, but in the end we will come out on top because we built a co-pilot engine on purpose.

No. 2, it should be transparent – ​​I usually say it’s a white box. I also use the terminology of reasoning. That’s an important criterion for us because you have to make sure that it’s super clear to all those lenders how you come to the conclusion, how you make the reasoning based on your interaction with the borrower, so that lenders can go back and trace 100% how everything is evaluated against their standard.

The final thing that sets our technology apart is how we protect consumers’ personal financial information. If a lot of companies say they’re using generative AI, okay, if they send consumers’ personal financial data to them Open AIthat’s not going to work. That’s why we purposely built our own library and decision engine so that we can protect that information and then prevent any kind of leakage by using a generated AI part.

SW: How are you using AI differently than some older companies?

DY: Our AI co-pilot Solo works through the consumer conversation, interacting with the consumer so that we understand the purpose of the consumer interaction and also get permission to access their data. When the consumer gives us access to their information – for example for the credit report – we would use our AI engine in the background to evaluate the individual’s credit report data.

We would identify the items from their credit report data that require further information based on the underwriting pipeline. So we would ask the consumer to provide those additional documents, knowing that once the consumer has submitted a mortgage application, an insurer will still have to ask these questions. We also calculate everything such as their income, and we also get permission to access their financial information about their income.

That’s the big difference from much older technology from lenders, which only collects documents to send to different departments. We’re already using our AI capabilities to communicate with the consumer, so once it gets to our customers’ underwriting department, they can look at it and check, check, check – everything’s fine.

SW: How does TidalWave fit into the next iteration of the mortgage industry, where there are fewer people?

DY: Lenders have already been cut to the bone, and they’re still losing money. And if rates rise, if volume returns, lenders have two choices. The first is to hire and bring back all those people as quickly as possible, and then train them in the process that people are not familiar with. So you start with a lot of problems, a lot of mistakes.

Or you take advantage of a new opportunity and help your existing employees to be a lot more productive; they can just focus on the things they specialize in and are so good at. And then you can control the volume when it’s low, and you can chase the extra volume when it comes back.

SW: What are your thoughts on safety?

DY: Coming from the background of building a technology platform, and then at FreeWheel, my first company, the type of company that uses FreeWheel’s technology is a company like NBC And CNNglobal media companies. So we understand 100% how important it is that the platform is very secure.

At TidalWave, we take the same approach because we understand that attracting customers who handle personal financial data can make or break a business if you do it wrong. That is why we pay attention to safety from day one. For example, even at this stage we encrypt everything so you don’t have to worry about being hacked and consumer information being leaked.

SW: What keeps you awake at night?

DY: My concern is that volume will come back and many lenders will have no choice but to run out and hire a lot of people, essentially having to go through that hiring/firing industry nightmare again. So as a small startup, we moved super fast because we want to make sure we can help as many lenders as possible so they don’t have to go through this nightmare.

SW: What does your team look like?

DY: We are a team of 15 and growing. Half of the people live in New York, so we come to the office four days a week. And for a small startup, I have to say that face-to-face is so important, especially because we are evolving so quickly.

This is my second company and I’m so lucky, especially from a tech talent perspective. I was able to bring the best team with me. My core team, co-founders, are the ones who built the first company from scratch with me. They followed me into my time at Better and then followed me out of Better, so that team has the benefit of working together for about 15 to 20 years.

And then the other half of the team consists of the members I had the pleasure of meeting during my Better Days. I also have people with over twenty years of experience in mortgage lending and loan advisors working with us. The advantage of a small team with the right level of expertise is that we work together for the same common goal.

SW: As a startup, how do you deal with the pace of change, especially in AI?

DY: You need to make sure you focus on what you specialize in. So we focus on the underwriting and we focus completely on the mortgage origination knowledge that we know no one else will build. Now we’re seeing a lot of companies building multiple large language models, which is a good thing because we can use them all, and whichever one is the winner, we’ll do deeper integration with it.

That’s what I think about famous quote about software eating the world, and now AI is eating software. And we see that right in front of us. With the help of AI capabilities, it can gradually take away all existing existing software capabilities piece by piece.