06-23-2022 SPECIAL EPISODE: TRUE Brand Launch With Stephen Epstein, Liz Long Adkins And Bob Noble

Facebook Twitter LinkedIn

In our Hot Topic this week, we have Stephen Epstein, CMO at TRUE amd Chief Marketing Officer at SoftWorks AI, Liz Long Adkins, Marketing Director at Softworks AI, and Bob Noble, SVP of Products and Innovation at Softworks AI to discuss with David how the launch of their rebranding is coming along.

Want to know more about Stephen Epstein?

As CMO at TRUE, Stephen is responsible for the firm’s go-to-market strategy and related functions, including branding, positioning, product marketing, and demand and lead generation.

Stephen brings over 20 years of marketing and product executive expertise and led these functions at industry leaders such as Mantas, Digital Reasoning, and Kasisto, with a deep knowledge of the financial services industry from his time at CDC Capital, JP Morgan, and Deutsche Bank. He has helped both fintech companies and global banking organizations flourish under his guidance.

Want to know more about Bob Noble?

As Senior VP of Product and Innovation, Bob is responsible for the direction and product roadmap of TRUE’s solutions. With over 30 years of experience in providing innovative technology for Fortune 250 financial services and mortgage companies, he is passionate about AI, process improvement, and automation, holding several patents in these areas.

Bob holds a Bachelor of Science in Mathematics and Computer Science from Stockton University, a Master of Science in Mathematics from Stanford University, and a Master of Science in Computer Engineering from North Carolina State University.

Want to know more about Liz Long Adkins?

Liz Long has over 9 years of experience working within the technology industry as a marketing professional. She now serves as the Marketing Director at TRUE, handling their omnichannel marketing campaigns driving brand awareness, product adoption, and customer retention. Prior to TRUE, Liz was the Senior Marketing Manager for the Americas at Thomson Reuters and the Global Marketing Manager for Digital Reasoning.

—

SPECIAL EPISODE: TRUE Brand Launch With Stephen Epstein, Liz Long Adkins And Bob Noble

I am honored and excited to have Liz and Stephen from what has been Softworks AI. We’re going to be talking about a rebranding and announcement. Stephen and Liz, thank you so much for being here.

Dave, thanks for having us.

It’s great to be here.

That’s good. You’ve got some exciting news so let’s get right into that. What is going on? What is the latest?

The big exciting news is that Software AI will now be known as TRUE. That is true and that is absolutely true.

Let’s get into the purpose of the rebranding and why this is so significant. You explained it to me on our pre-call. I’m excited for our audience to learn more about why TRUE is such an accurate description, the name and a rebranding for this product.

It’s an interesting story as we discussed. The company was started back in 2017. One of our cofounders, Dr. Ari Gross, has been focused for quite some time on the notion of applying AI and machine learning to the areas of character optical recognition, data extraction, and so on and so forth. Over the years, as software started to grow and a variety of lenders started to use our technology, it became pretty clear that the underlying AI models, the capabilities, and the machine learning that was being used around the data science in the company was starting to produce amazing results.

Like any data science and AI company, we take all the results that we have, and we’re continually using that to retrain and reinforce our technology to make it smarter. A big part of what our AI lab does here is essentially that. It’s always retraining and reinforcing our models. As our technology got smarter and more and more customers started using it, at this point, 50% of the mortgage insurers in the industry use our technology.

That’s a significant market share.

It is. With all this data coming in and technology being retrained, reinforced and improved, it was becoming very clear to us that the underlying product was delivering truly amazing results to the industry. There’s always a lot of skepticism around AI and machine learning in any industry, not only our own industry, but it was incredible because we sit down with customers and understand exactly how they’re applying our technology and the results we’re driving.

Through that, it became clear to us that we need to be clear to the industry in terms of what our value is and why we’re different from all the other AI platforms, data extraction utilities, and verification utilities. We started through this journey over the past few months to basically unpack and start to develop that. What became clear to us was that at the end of the day, we are delivering true results.

I talked about mortgage insurers. One of our customers improved underwriter performance by 300% to 400%. That was a crazy improvement. When we were done with our messaging and our positioning here in marketing and being bold about how we talk about the company, we decided to be honest with ourselves and the industry and to come up with a brand that said how we feel and what we think we do. It centered around this notion of TRUE and being truthful, getting truth out of the data, and getting a truthful, holistic, unbiased view of lenders. It all comes together. That became the power behind our brand.

Brand Launch: The notion of being truthful, holistic, and unbiased becomes the power behind our brand.

It’s very exciting. Joining me on the interview here is Jack Nunnery. Jack, it’s good to have you joining in on this exciting interview. I appreciate you being here.

Stephen, how are you?

Good, Jack. How are you doing?

Good.

One of the things that I like in the rebranding and as the Chief Marketing Officers, CMO, there, I was intrigued with the design of the logo. A lot of times, people go find a graphic designer and they put something that looks cool, but that’s not the story. In this logo, you have a very unique reason why you created it. Explain that to our audience, please.

The logo represents our journey, what we’ve learned, and what we do. You said it well, David. It’s not a typical logo where we set a designer in a corner and said, “Come up with something cool. If we like it, we’ll go with it.” We sat down and talked about our journey, what we do, and where we’re going. The logo represents three things. When you look at it from left to right, you see these lines coming into the logo. They represent all the different types of data that our customers are continually bringing in. Things from W-2s to pay stubs to house valuations, whatever the data might be. A lot of data streams are coming in.

As the data comes through our platform and as you envision it in our logo, you start to see all these streams coming in. The middle of our logo essentially is an eye because what we do at the end of the day is look at every single piece of data coming in and every single data point. We extrapolate what’s real, what’s not real, what’s true, and what’s false. Sometimes, data is not always clear. You can get a scanned image that’s scanned crookedly. It’s not clear or fuzzy. We’re always looking at all the different data points. We’re comparing it to other documents in the loan pool and we’re figuring out, “Maybe this is not an eight. Maybe it is a B because it’s represented as a B in another document.”

The eye represents the fact that we’re looking at everything and applying our intelligence to figure out what everything means. When you go all the way to the right of the logo, you’re left with a point. The point is that this is the data insight you need as a lender to understand what your borrower is all about. The logo is simple. We’re taking a lot of data in, analyzing all the data, applying intelligence to it, and showing you the insights that help to power your automation process and all your lending processes.

Stephen mentioned that their approach is to look at data as it presents itself across many documents and the loan file. What looked like an eight on one document is confirmed to be the alpha character B by five other documents. At least in my experience, I find this to be a very good practice because data replicates itself through the loan file. I did an analysis once, Stephen, that in the average loan file, it’s always about the average loan file. There are generally 3,000 pieces of critical data that are contained inside the average loan file. To be able to use those to correlate and increase confidence in the value that’s being extracted for utilization in any type of robotics or AI seems to me to be a big win.

It’s a tremendous point you’re making. At the end of the day, if you think about this process when it’s human-driven, we only have a certain amount of fidelity in our brains. We can only remember a certain amount of information at any given time. If you’re looking at 100 files and 3,000 data points, you may say to yourself if you’re a seasoned underwriter, “I remember I saw that somewhere.” You go and find that document, and you say, “This data doesn’t look right because I saw that.” That’s an outlier. That’s not the norm. When you think about a machine, a machine can remember everything.

It’s such a lost opportunity when other technologies don’t look at all the data across every single document in every single field because once you have that and you have the intelligence to understand what it means, you can do so many things. I use the B’s and the 8’s as an example. You can turn it upside down, and you can actually use that to validate someone’s income. Is this the same income that is showing up on pay stubs, tax records, or bank statements? You can review every single one of those. You can uncover fraud if you have this holistic view of all these documents and data points. That’s a big part of what we do, the platform, and our underlying products.

Brand Launch: Once you have the technology to look at all the data across every single document in every field and have the intelligence to understand what it means, you can do so many things.

Again, it comes out so graphically well in the logo. You talk about making smart lending decisions. Both of you can respond to this, but how much of the AI factor is in the smart lending part of this?

I learned this from one of our data scientists quite recently. A good example of this is if you think about automation and a lot of the platforms out there that are automating all these different tasks that take their lending process, data plays an incredibly big role obviously in that. If your data is off by 5%, the accuracy and the overall lending process or the automation process can be off by as much as 30%.

The individual going through the automation, the underwriter or the knowledge worker, their workload may increase by as much as 30% to figure out what’s wrong in that 5% of data and how it affected their decision. Think about that. If it takes them 30% longer to figure out what went wrong in 5% of the data, it’s going to take them longer to make a decision to turn it upside down. The cleaner the data, the less they have to figure out what’s right and wrong and the quicker they will make a well-informed decision.

Liz, do you want to add to that?

A quick decision as well always comes to play. Another example we always bring up is that you can go and buy a Lamborghini in an hour and have a loan ready for you. That can’t happen for the mortgage industry right now. The ease and quickness that a decision can be made are also important.

Jack, do you have a thought you want to add on?

I go back to increasing underwriting efficiency by 300% or 400%. Stephen mentioned that number earlier in the discussion. I was sitting here thinking about the methodology, David, that we use, standing up a corresponding aggregation business where we looked at all of the milestones in the process, all the jobs associated with the milestones, and then ultimately drilled all the way down, Liz and Stephen, to the task level.

As you think about it, you’ve got to do these 12 tasks to accomplish this 1 job. You have to do these nine jobs to accomplish this milestone. The most expensive, whether or not it was a job task or milestone, was in the area of underwriting because you spend a lot of time and the people that perform that function are some of your highest-paid associates. We call them actors in this particular analysis.

You’re talking about a significant usage of time by associates that are among your highest-cost operational associates. I’m not comparing them to loan officers out there, but inside of your operational fulfillment center. What this can do is significantly reduce the cost to originate, which David and I pegged at $10,637.

The question out in the marketplace now is, “How do we not stop the increase in cost but reverse it pointed in a direction going back down?”A 300% to 400% pickup and underwriting efficiency is a huge needle mover on a profile basis. Jack, that data came from Bob Noble, who’s our SVP of Product. He recently joined us from Genworth, and he was a customer of ours. He applied the technology to the underwriting process there.

Bob, it’s great to meet you. What’s so interesting is your story and how you connected and decided to join up with TRUE. What’s your background and how did you discover all this?

David, thanks a lot. I have a deep background in the mortgage space for many years in technology. I started out with GE and worked through many capital businesses in the mortgage space. One thing we learned over time is that automation is key to driving efficiencies and accuracy. The key to that is data. Data is king. A few years ago, when we were trying to do analysis and work on our technology modernization, we looked at many tools to figure out how we could turn unusable and unattainable data from documents into real data that we could use in our underwriting systems.

Automation is key to driving efficiencies and accuracy, and the key to that is data. Data is king. Click To Tweet

After a long bake-off, we came across Software AI, soon to be TRUE. They rose to the top of the list. Implementing this technology, we immediately saw the benefits from an operational standpoint. Prior to this tool, it was a very laborious task to data enter loans from these voluminous loan packages of 250 pages and hundreds of documents. To get the data fields that we needed for our underwrite, we tried to slim it down to only a few hundred was taking us over an hour to do that.

We had a team of over 25 people doing nothing but clacking in data and trying to index a loan. Because it was a manual process, our accuracy wasn’t fantastic either. Fast forward after the introduction of this tool, we were able to reduce the number of folks to about two and a half. That was during the biggest boom in our business. We were able to classify over 150 documents and extract over 3,000 data elements. It’s impressive.

The key is that not only was it faster and cheaper. It was better. The data was accurate and it was correct. We had confidence in it. Once we had confidence in this data, we were able to use that in many different areas. It was a great tool for analytics to be able to re-look at your guidelines and your potential and find ways to mitigate risks alone.

It also allowed us to automate processes downstream. Because we had trust in data and a lot of great rich correct data, we were able to automate and build models that will allow us to fast-track some of our loans. We saw a jump in our underwriter productivity from a 3X increase. Our underwriters became more efficient. Ultimately, our customers were happier because they were giving their loans back quicker. We felt confident in our ability to accurately underwrite these loans because of our automation and data, and we had consistency in our process. It was a win-win all the way around, better loans at a lower price, better for our customers and business.

You increased underwriter productivity from three loans a day. If it’s a 3X, that would be 9.

It was about 3 1/2 loans to just under 13.

That added credibility to you. The fact that you downplayed it. That’s always the key to it. That’s interesting and significant. Now I want to put this in context. This is why you were working at a mortgage insurance company doing contract underwriting, I assume.

We did contract underwriting in mortgage insurance sometimes. We did both guideline underwriting and risk underwriting as well as QC. It was a full gambit of things.

That was a significant impact. Bob, thank you so much. I appreciate you injecting that in here.

My pleasure. That was the outcome they saw. When we were fully implemented, they were seeing that 300% to 400% increase in underwriting productivity. On the flip side of that, you mentioned cost savings. They were spending quite a bit of money on data entry and data correctness and reduced the amount of data correctness. Overhead, they needed to accomplish this early on in the process.

They brought it down from 25 data entry clerks down to 3, which was an important moment for them in terms of not only reducing cost but also creating that clean, efficient data coming in, which then ultimately led the underwriters to have a much more efficient process that boosted their productivity. They lowered their data entry costs because the machine was processing data much more efficiently and accurately. They increased their productivity because the data was that much cleaner when it came to the data underwriting process.

Brand Launch: Creating that clean, efficient data coming in ultimately leads the underwriters to have a much more efficient process that boosts their productivity.

Does the machine or technology call out the confidence level in a particular data element? Fine. You’re going to go from 25 to 3 data input people. What that means to me is that somehow the machine is telling me, “I have a 98% or 99% confidence that this eight was the alpha character B,” and then you pass that as a valid piece of data wherein something may have a 70% confidence rating and then that kicks out to human intervention to get eyes on the data field. Is that how this is working behind the scenes?

That’s essentially how it does operate. Of course, every prediction it makes in terms of data outcome has an accuracy and a confidence score against it. The customer can determine where they want to start, pulling humans into that process. It’s a threshold. They can set the threshold low early on. Many of our customers do so they can see the accuracy, and then, as they get their own confidence, they move that up. Some of our customers move that threshold up as high as 95% to 99%, depending on where they are in the implementation of the product. Every prediction the platform makes around the data it processes, there’s obviously a confidence score against it.

Can you vary the confidence score by data field? For example, the subject property address.

By data field, absolutely. I can segment those, too.

David, that’s helpful when you can push that confidence factor before human intervention up on a data field that is critical to a successful onboarding into your servicing platform. I’m thinking downstream now in the process. One thing that is always painful is getting a rejected servicing boarding file because you only have three days to get those corrected, and all of a sudden, it’s a calamity. If you can push the confidence scores higher on critical fields at different parts of the process, and the first one that popped into my mind was servicing onboarding fields, then that gives the user a lot of power in terms of managing time spent with a human validating data.

Jack, that’s an important point. Our goal with the technology here is to make the knowledge worker that much more productive, so they can focus on being a knowledge worker. In this case, an underwriter being a risk manager instead of an underwriter being a data verification clerk. Let’s give the machine these problems to solve to free up people to do the high-value tasks they’re hired to do. That’s a big part of what this thing does.

Our goal with the technology is to make the knowledge worker that much more productive, so they can focus on being a knowledge worker. Click To Tweet

One of the things I like about your new logo, first of all, the name implies what you’re doing. It brings true data forward, helping people drive costs lower and with greater accuracy. All those things come from having empirical data. Also, you talk about lending intelligence. Below your new logo, there are the words lending intelligence. Talk about that and what that means to anyone tuning in to this show.

It’s a simple idea. Through our journey, speaking with so many customers, the notion of truth and data would surface all the time. We’re an AI company focused exclusively on the lending industry and everyone’s telling us, “You’re letting us understand the truth in all this data.” You’re truly an intelligent technology in the lending industry. We started to get centered around this notion that if truth is in data and lending data gives you the truth about your borrowers, let’s focus all our positioning around lending intelligence. As a lender, you can have true intelligence about everything you do and who you lend to through lending intelligence and through our TRUE platform.

Liz, do you want to jump in with some comments behind that?

Our lending intelligence is able to ensure that all the data in the full automation process is correct from the beginning. That allows smart decisions to be made throughout the process. That’s what sets us apart and that’s also how we are able to plug into different other automation systems to make the loan process truly touchless from start to end.

It’s important that we stress the point that you are not a decision technology. What you are that is you feed true data into whatever system they’re using. Expound on that a little bit if you could.

At the end of the day, our mission here is to power all the LOS systems and lending automation systems out there by ensuring the data is going in. It is super accurate and highly organized, with a lot of analysis and intelligence behind it. We can effectively power that whole automation process. By the way, we’re talking a lot about the process of getting data in.

Once the data goes in and the decisions are made, there’s then an opportunity to use the same technology to go back and retrospectively say, “Was the right data used to make the right decision, or are there some mistakes in this?” We have a product called Verify that our customers use to go back in and verify the loan pools using this technology to uncover if there are errors in any of this, which helps to mitigate risk.

Once the data goes in and the decisions are made, there's an opportunity to use the same technology to go back. Click To Tweet

I was wondering how Verify plays into the equation because you’re extracting at confidence levels that are user-set, giving you a TRUE data set. Explain to me why is there a need for the Verify piece.

Not all the data is coming through our channels. In the automation process, a lender may pull in data from other channels. If they pull in structure data, you’re left with a loan package or a loan pool that has data from a bunch of sources. Don’t you want to go back in and revalidate that to make sure it’s what it needs to be?

Conceptually, you take a data dump out of the entire file, run it through Verify, and then look for inconsistencies between all data in the file using what you did as the baseline, and then compare other that was verified or validated through you against the baseline data, which you ran through your extraction protocol.

That’s precisely what we do, Jack, and I couldn’t say it any better.

I love the new logo and the new look. You’ve been doing this for a while. It’s not like it’s brand-new technology. We’re not launching a new product. We’re just launching a new branding. It’s important to understand that this is a company that has existed. Again, you’ve talked about what Ari started, how he created this thing and where it’s going. Where’s this all heading? Where do you see TRUE going as far as it relates? What can we anticipate moving forward in the months and years ahead?

As I said earlier, our Founder, Dr. Ari Gross’ vision is very simple. He wants to see and we all want to see the same thing in real-time loan decisioning in the mortgage industry. A big part of this is all about data. The automation platforms are very good. If they have great data coming in, there’s validation, and it’s all operating in real-time where machines are learning. There’s no reason we can’t move this industry to a touchless real-time experience. That’s a big part of our mission.

Brand Launch: If great data is coming in, there’s great validation, and it’s all operating in real time where machines are learning, there’s no reason we can’t move this industry to a touchless real-time experience.

Kudos to you and Liz for putting together what is an amazing logo for an amazing company and product that couldn’t be more timely. Anytime we could improve productivity by 300% to 400%, it’s going to have an immediate cost-cutting component. How could people go and learn more?

Softwares AI will be redirected to the new brand and the new website, but the fastest and most efficient way to get to us is www.True.ai.

Thank you so much for taking the time. This is exciting news. I am very excited about this because so many of our clients and so many in the industry are looking for true data. We don’t have it at near the levels we need to, and therefore the accuracy is a problem. It’s so good to have both of you on. Jack, I appreciate you joining me on this interview. I’m excited about this product.

Again, what can it do for our industry? It’s increasing efficiencies and productivity by 300% to 400% and knowing that 50% of the MI companies are already using this product. Folks, you got to pay attention to this. Forward this show and interview to those inside your company that would benefit from this knowledge. Thank you, Stephen and Liz, for being on the show. Be sure to say hi to Ari. I enjoyed meeting and talking to him.

Thank you.

About Stephen Epstein

As CMO at TRUE, Stephen is responsible for the firm’s go-to-market strategy and related functions, including branding, positioning, product marketing, and demand and lead generation.

SPECIAL EPISODE: TRUE Brand Launch With Stephen Epstein, Liz Long Adkins And Bob Noble

Important Links

About Stephen Epstein

About Bob Noble

About Liz Long Adkins