A myriad of documentation is needed for lenders to make an accurate lending decision. This becomes even more complicated in the case of auto loans where an entirely new category of paperwork enters the picture. Lenders need to be able to process this unstructured data quickly and easily.
Our next guest on the Fintech One-on-One podcast is Justin Wickett, the CEO and Co-Founder of Informed.iq. They have tackled this challenge head-on and are able to automate verifications in real-time with 99% accuracy.
In this podcast you will learn:
- How Justin’s time at Lyft and Credit Karma helped solidify the idea for Informed.
- Details of the Informed core product.
- How the automated verification process works.
- How the gig economy has complicated income verification.
- The kinds of documents they are processing.
- How they are using AI and Robotic Process Automation.
- What they are doing when it comes to fraud prevention.
- Why they decided to start with auto loans.
- The core target market for Informed.
- What their partnership with Origence (formerly CU Direct) means.
- How they work with credit unions, where there are limited tech capabilities.
- Justin’s vision for the future of Informed.
You can subscribe to the Fintech One on One Podcast via Apple Podcasts or Spotify. To listen to this podcast episode, there is an audio player directly above or you can download the MP3 file here.
Download a PDF of the Transcription or Read it Below
FINTECH ONE-ON-ONE PODCAST 380-JUSTIN WICKETT
Welcome to the Fintech One-on-One Podcast, Episode No. 380. This is your host, Peter Renton, Chairman & Co-Founder of Fintech Nexus.
(music)
Before we get started, I want to remind you about our comprehensive news service. Fintech Nexus News, not only covers the biggest fintech news stories, our daily newsletter delivers the ten most important fintech stories into your Inbox every morning and we have special editions for Latin America as well as UK and Europe. Stay on top of fintech news by subscribing at www.fintechnexus.com/subscribe
Peter Renton: Today on the show, I’m delighted to welcome Justin Wickett, he is the CEO and Co-Founder of Informed.IQ. Now, Informed are a very interesting company, they are focused on document automation, document analysis. When you go to apply for any kind of loan, they focus specifically on the auto space primarily, at least right now, so many documents are needed and the lender has to process all these documents and to do it efficiently, you really don’t want to do that with a manual review, you want to be doing this in an automated fashion and Informed, they provide the technology to do that.
So, we talk about obviously that technology in some depth, we talk about the different ways they do it, we talk about the different types of lenders they work with, we talk about fraud detection. We do a little bit of a deep dive into credit unions because they’ve just signed a big deal there and credit unions have unique needs in the space and so that’s a really interesting piece of the conversation and we talk about what’s next. It was a fascinating interview, hope you enjoy the show.
Welcome to the podcast, Justin!
Justin Wickett: Peter, I’m delighted to be here, thanks so much for having me.
Peter: My pleasure. So, let’s get started by giving the listeners a little bit of background about yourself. You’ve done some interesting things in your career, why don’t you give us some of the highlights before Informed.
Justin: I have a background in engineering, I did computer science at Duke University and afterwards went into product management, I have been a product management leader most recently at Credit Karma, they provide free credit scores and credit reports to over a hundred million Americans. Before that, I was early on at Lyft, the ride-sharing company, I was responsible for passenger acquisition and engagement and then before that, I was at Zynga, the social gaming company. Actually had a lot of exposure to fraud and anti-money laundering, unfortunately, during my time working for Zynga Poker, the world’s largest online free-to-play poker game with over 35 million players every month. So, that’s my background.
Peter: Well, that’s such some great names that have been a part of there. So, don’t you then tell us a little bit about the founding story for Informed. What was the impetus to get that off the ground?
Justin: It all builds up. I was at Lyft and I got to see my counterpart responsible for driver acquisition and engagement, really struggled because so many of the people signing up to drive on the Lyft platform, they didn’t have a vehicle that could qualify and they needed to go out and get a loan to purchase a car. I got to see during that process how broken the underwriting process was for a lot of these people, a lot of people don’t qualify for the 0% APR that you see and the leases that you see on television ads. These people would spend the whole day at the car dealership trying to qualify for a loan and they weren’t sure if the lender was going to take into consideration their overtime pay or their bonuses or their tips and it was a broken process.
So, I actually ended up getting involved with Credit Karma because I really wanted to deepen my relationship with financial institutions who were lending money, who had a lot of experience doing that to understand what their pain points were. I saw how financial institutions really wanted to strengthen their own brand, they didn’t want to just prop up the Credit Karma brand, they actually lacked great software solutions to drive the conversion.
You’d fill out an online application and after submitting the application there’ll be like a 1-800 number that you were instructed to call to get an update on your loan or it would take days to get a call back in terms of what the status of your loan was. So, I knew that if we could craft software to ultimately help strengthen the financial institutions’ brand, provide for real-time decision making, more transparency and improve access to capital and financial inclusion, we would be in a great position. So, that’s why we started Informed.IQ.
Peter: Okay. So then, maybe you could just describe the product, now the core product, how does it work? I’d love to kind of get a pretty solid description here.
Justin: Informed automates verifications for over 150,000 Americans each month by turning documents and data into decisions in seconds using machine learning and Artificial Intelligence without having to rely on people that are notorious for introducing bias, for making mistakes. So, that’s the product that we’ve got, it’s APIs to automate the back office of a bank and we typically integrate with the loan origination system which we kind of think of as Craigslist, if you will, it’s trying to do so much, but it doesn’t do anything particularly well. So, we’re the AI that is replacing the verification screen within the loan origination system.
Peter: Okay. So, let’s dig into that a little bit. Can you maybe provide an example of some of the documents you’re talking about there, the data you’re talking about there, how you process it. Just take us through kind of how it works.
Justin: This is really what fires up our team because the company, even though we’re a B2B company, we’re very mission driven. We talk about in our all-hands meetings, examples of how a car dealer was trying to get paid by a bank, like Capital One, for some kind of warranty product or insurance product that, in fact, the car buyer never even agreed to, they never even signed for that yet the car dealer is trying to get paid.
So, our software is capable of going through complex contracts that a lender would receive from one of the 30,000 dealers across the United States, documents like retail installment sales contracts, ancillary product contracts like vehicle service contracts, pre-paid maintenance plans, tire & wheel contracts, nitrogen tire-filled contracts. There’s over 8,500 different variations of these contracts that are out there that car dealers can sell, not to mention vehicle valuation guides, hook out sheets or title documents and odometer statements.
So, Informed processes all sorts of different documents and turns them into data in a matter of seconds so that lenders can lend with greater compliance, they can avoid consent decrees, they can better adhere to the policies and procedures that they’re representing to their regulators and the rating agencies that are in the securitization process and not to mention income documents. We process tens of millions of pay stubs and Social Security and award letters and bank statements and W2s and 1099s, the list goes on and on and on. And what fires up our team is about 30% of the people applying for credit in the United States, they actually understate their income, especially we see this trend in the non-prime consumer base, people understate their income and the question is, why are you understating your income, you’re going to end up with a higher interest rate, you’re going to spend thousands of additional dollars on your loan repaying that interest, why would someone do that.
The reality is we dug in as we’ve conducted these user interviews and talked to folks, they just don’t know. If I’m an hourly worker and I’m applying for credit, I don’t know how Wells Fargo is going to interpret my overtime pay or my double time pay or my commissions or bonuses. I don’t know if I should include that or not include that and so it’s confusing and as a result, we see instances where a car dealer will actually ratchet up someone’s income.
A person will walk in and say look, I make $50,000 a year and the car dealer will submit the credit application on their behalf stating that they actually make $70,000 a year and they get stuck in a car that they actually really can’t afford to make the payments on just because the dealer wanted to sell the vehicle and vice versa. So, we see dealers that are struggling to figure out, how is Wels Fargo going to take into consideration this person’s commissions and overtime pay and bonuses and they might understate the applicant’s income. As a result, they might not be able to help that individual finance a vehicle that they really need.
Peter: So, let’s stay on income for a second because, obviously, you mentioned time at Lyft and, obviously, lots of people have a regular job and they drive for Lyft on weekends, nights, how do you take that into consideration and how is that presented as income?
Justin: The reality is there needs to be a lot more transparency in the space. So, we actually see the CFPB putting out an advance notice to propose rule making in ANPR related to Section 1033 of the Dodd-Frank Act which mandates that financial institutions need to share with American consumers the data that was used to render a financial decision. What we see in the industry is more and more Americans are earning income from a variety of different sources. A lot of people are not just W2 wage earners, they might have W2 wages, but they might have some other form of fixed income like supplemental security income or social security income or they might be driving for Uber and Lyft or renting out a bedroom on Airbnb so they’ve got additional forms of income.
What we found is that banks are notoriously poor at trying to add up all those different forms of income. A lot of these lenders will have to hire up temporary staff and train these folks to be able to do these, people don’t have the experience, sometimes they try to offshore and that results in a disaster where people’s incomes get calculated differently and they don’t necessarily get the most fair rates. So, Informed has spent years processing tens and millions of these bank statements, automating the extraction of line items and deposits.
How can you solve really complex use cases, like what if the bank account is a joint bank account and there’s actually multiple bank accounts on a single bank statement, what if the bank account is a personal bank account versus a business bank account, what if it’s held by an individual versus jointly held. So, there’s all sorts of use cases that need to be taken into consideration to ensure the accurate calculation of an individual’s income.
Peter: That’s still a complex problem. So then, I want to talk about document process automation you talk about it on your website and can you just describe it for me and like there’s two parties here or there’s really three, you’ve got the dealer, the lender and the buyer of the car, are you talking about all three of those kinds of entities or what kind of documents are you analyzing?
Justin: So, we started with documents because that’s really where the pain was in the industry. The bulk of the lenders that we serve, they are receiving literally hundreds of millions of these documents from all of these dealers across the United States and the American consumers uploading documents because let’s face it, if I’m a construction worker, if I’m a painter, I can’t sign into some online solution and provide for my pay stubs to be pulled down. It doesn’t exist, they don’t support my construction company yet.
Peter: Right.
Justin: And those are the people that really need this type of fair income treatment the most. So, Informed started off with documents, we’ve been layering in with our partnership with Truework and Plaid and Finicity and Troove access to credentialed income data sources as well, but for the most part, we knew that we needed to be very good at lifting data off of documents in an unbiased, automated real-time manner to facilitate for the transparency that Section 1033 of Dodd-Frank mandates.
We knew that, ultimately, financial institutions, part of the reason why credit is so costly for consumers that are non-prime, that need their information verified, that have a lot of verifications is because there’s a lot of manual reviewer that needs to take place. If we could use software to automate that manual review, to create a more fair, streamlined, more transparent process we could lower the cost of credit in the United States.
Peter: Right, that makes sense. So then, what about AI, robotic process automation, you know, how are you using those technologies?
Justin: We got started because a lender handed to us a whole bunch of PDFs and said look, we are getting these 100-page PDFs that were scanned in or faxed on over and we’re spending days, it takes us days, if not weeks, to get one of these car dealers paid because we’re having to go through and verify that the signatures are all in the right places, that the initials are in the right places, that the numbers all add up, it just takes forever. There’s always defects and compliance issues that pop up so can we automate that. And we actually tried to use Google Could platform, their OCR solutions, AWS, their OCR solutions and we realized that there wasn’t going to be enough, we needed to build a lot of intellectual property on top of that data to really be able to deliver against the service level agreements, the accuracy, the precision and recall that these lenders required. To be able to undergo the model risk process mandated by the OCC really required a whole different level of scrutiny and investment than just what a vanilla OCR solution from Google Cloud platform or AWS was offering.
So, we went ahead and begun to build the underlying machine learning models on top of OCR data to classify documents, to extract entities, to do the appropriate comparisons and validations and not just OCR on a one-off basis, in a stateless manner, but really compare the results to all the historical loan jackets that Informed has ever processed, all the historical pay stubs that we’ve ever processed. So, if you think about it, OCR just reads information off of a document and it’s stateless, but Informed knows hey, a cashier working at Walmart in Fresno, California tends to make this much in overtime, this much in terms of bonus and so we know what ranges are acceptable and we can cross validate against that.
Peter: Right, right, got it, okay. What about fraud because I’m curious, is that really a part of what you guys are offering here because, I imagine that in all types of lending situations there are attempts of fraud. How are you kind of tackling that?
Justin: Yes. Fraud is a very costly problem and it’s on the rise, especially in light of digital retailing and more lending moving online outside of the bank branches so we knew from day one that we needed a solution to fraud. Fraud is a complex problem because there’s disclosure requirements on the financial institution so Informed actually goes way beyond just document process automation, we actually, in addition to reading information off the documents, go validate it against information found on the worldwide web. So, if you go to Google and search fake pay stub generator there’s like over 100,000 different websites, listings that come back and allow you to print out fake pay stubs, fake W2s, fake bank statements.
It’s crazy the things that banks have to rely on people offshore to memorize all of these different forms of fraudulent pay stub templates out there, it just doesn’t work, it doesn’t scale, especially in a cyclical business where lending is a fix up during the tax refund season, it doesn’t work. So, that’s why lenders have been so excited about embracing AI technologies from Informed that can enable them to board loans in a real-time manner, in a compliant manner that helped to identify fraud that doesn’t get caught. We actually were doing a quarterly business review with one of our large customers and were talking about Informed having identified over $10 Million worth of fraud that we helped to protect against that otherwise would have gone towards their loss ratio.
Peter: Interesting, interesting. So then, we’ve been talking primarily about the auto space, but obviously the problems exist in most lending verticals, the problems that you guys are addressing, are you focused primarily on auto or what verticals do you work with?
Justin: We started in auto given my background, coming from Lyft, coming from Credit Karma, auto is a massive vertical, there’s 35 million indirect auto loans originated in the United States, contrast that with about 8 million mortgages, obviously, mortgages is a higher ticket item, but there’s a lot of documents and data associated with an auto loan and auto is very democratic. The American population, everyone needs to buy a car, the person that maybe is your housekeeper or your gardener, they need to have a car to get to work and frankly, we knew that if we wanted to lower the cost of credit in the United States, if we wanted to improve financial inclusivity and create a more fair financial system, we knew that we needed to actually go to where the problem was which is the bias that exists in auto lending so that’s where we started.
Now, you raise a really good point, which is all of the tens of millions of pay stubs and income documents that we process today, those are the same pay stubs that are getting uploaded when consumers want to refinance their credit card debt with a personal loan or they want to apply for a HELOC and tap into the equity that they gave built up in their home. So, Informed is working with lenders like Avant, SoFi and others to provide for a more real-time, more transparent and fair income calculation abilities, fraud detection abilities so that they can better serve the American consumers who are trying to figure out their financial options.
Peter: Right, right. So, that’s obviously some of the fintech lenders, I mean, I presume you work with traditional lenders and I know we’re going to talk about credit unions in just a minute. but what’s the core target for Informed.
Justin: Informed started off with big banks. We actually tried to go after, maybe the most challenging customer segment for an early-stage startup, why would a big bank want to share any data with an early stage startup. But we knew that the big banks and credit unions and state-licensed finance companies, they were the ones that were ultimately serving the broadest population that really needed this kind of automation so we wanted to start there.
We have a lend and expand model where we roll out in say the auto division of a financial institution, we have them test out the software, they run it head-to-head, they do a champion challenger assessment with their existing manual funding process and they realized, oh my gosh, this is so much more accurate, it’s incredible. When we get an email from the senior vice-president of lending at a big consumer bank saying that the AI is actually more accurate than their own staff was at calculating people’s incomes and so that enables us, once we get these kinds of testimonials, that enables us to expand to other divisions of the bank like their unsecured personal lending division or HELOC division and that’s been very successful.
Peter: Right, right, okay. So, I do want to dig into the credit unions for a bit because I think it was a couple of weeks ago that you guys announced you’re partnering with Origence and like you can explain who they are and what was involved in this partnership.
Justin: We’re so excited about this, it really is the next chapter of our mission. Credit unions in the United States are not for-profit entities, they do wonders in terms of making capital available to American consumers promoting financial inclusivity so the opportunity to partner with Origence, formerly known as CU Direct, and serve over 1,100 credit unions and 15,000 dealerships in their network is incredible for Informed. Credit unions have two challenges that they talk about, one is they want more automated underwriting. A lot of the underwriting done at credit unions today is still manual, there are still loan officers that have to manually review income documents and manually review even credit reports, consumer reports to make an underwriting decision.
The second factor is credit unions want to speed up the funding of their loans, it takes them weeks on average to get loans funded because of the manual review that I spoke of in order to originate loans in a NCRA compliant manner. So, Informed is the perfect AI partner with Origence to speed up loan funding, to improve transparency, to catch critical defects in that origination process. And, yeah, we couldn’t be more excited about getting going with them this year and ramping in 2023.
Peter: It’s interesting because obviously there’s very different capabilities. When you talk about big banks or even companies like Avant and SoFi that you mentioned have significant tech capabilities whereas your typical credit union doesn’t. You know, they all would have some kind of loan origination system, I imagine, but I imagine they’re all different, right, or many of them are different so how do you work with credit unions that might have just a handful of people on staff who have any kind of tech expertise at all. So, how do you do that?
Justin: Well, that’s why the Origence partnership is so unique and strategic in the industry. Origence serves as the rails, it is the rails through which automobile dealerships across the United States submit credit applications to credit unions and receive credit decisions. So, given that we are integrated into Origence and an exclusive partner in this capacity, we are able to feed in a standardized way in all of the different credit union loan originations systems the data that is required in order to render a more fair, automated underwriting decision and loan funding decision. So, it is a very strategic partnership in that standpoint because we are right in on top of the existing rails that are well established in the industry.
Peter: Right, right, okay. So, when it comes to the credit union space you talked about the needs they have which every lender has similar needs, some of them just have just more capability, I mean, this was just announced a couple of weeks ago, have you started pilots with credit unions? I mean, how are you finding it different and how are they the same to other clients you work with?
Justin: So, we are getting swamped by demand, it is incredible how this really resonates with the credit union market. First off, credit union auto loan originations are at an all-time high, credit unions are very much focused on closing the gap and enabling for fair access to credit in this segment so credit unions have been very adamant about wanting to apply automation and speed up the loan funding process.
To your point earlier, it’s incredible that we get to take the technology that we’ve crafted in partnership with a Capital One Auto Finance, an Ally Financial or a Westlake Financial bring all those learnings as it relates to model risk management, all of the tens of millions of documents that we’ve been able to automate today and apply that to the credit union space and give them the expertise that we built up so we are very excited about that. Now, credit unions have some unique challenges that we’re going to need to overcome and we’ve been working with numerous credit unions, so far, helping to facilitate this one.
Credit unions have what’s called a membership application, you need to actually be a member of the credit union and fill out a membership application to prove that you meet the membership eligibility requirements which is effectively a stipulation on a loan. So, how can we help to automate that a particular American consumer is indeed qualified to be a member of a credit union that is offering a very competitive rate as a not-for-profit institution. So, Informed is very much focused on taking the machine learning models that we have a lot of success with, having the parameters, applying those to the documents that credit unions have their membership eligibility forms and automating those verifications to further lower the cost of credit and provide for more transparent, real-time decision making.
Peter: So, they’re looking at this not just to serve their existing customers, it sounds like they’re looking to this to really grow their membership base, that’s what you’re saying, right?
Justin: Absolutely, yeah, yeah. We see credit unions, again, being able to serve a need in the market as a not-for-profit institution, they are leveraging Informed not just to better serve their existing members, but to make it easier for new members to gain quicker insight into their membership eligibility status and have a more real-time loan boarding process along the lines of what Capital One and Ally, what other industry leaders have really pioneered and driven for.
Peter: Right, right, okay, So, maybe we’ll close with where you’re taking this, I mean, what’s next for Informed and where are you kind of…what’s your vision for the future?
Justin: We raised a $20 Million Series A, we have incredible investors, Nyca and US Venture Partners have been incredible support to Informed. We’ve got a phenomenal team and we absolutely have product market fit. In fact, we’ve got incredible demand from the industry to apply automation, to drive financial inclusivity, the real-time transparency and improved compliance around loan originations. Documents is just the starting point, it’s where we had to start to be able to deliver real value to our customers and to help Americans get access to low-cost credit at the best rates possible.
Where Informed really wants to go is leverage this contributory database that we’ve been building to make for more informed financial decisions, we don’t want to just be limited documents, we’re integrating and processing data from consumer credential data sources and really trying to remove a lot of the bias from the origination process. So, we’re very excited about that, we ultimately believe that we will be able to expand beyond consumer lending and provide the same level of automation to the 70 million Americans who apply each year for Medicaid, government-subsidized health insurance for the 40 million Americans who apply each year for supplemental nutrition assistance programs or low income on energy assistance programs.
All those people have to upload pay stubs, upload utility bills, prove that they qualify for this government assistance and again, it’s a very inefficient process today, there’s a lot of manual review. If we can use our know how, all these pre-trained machine learning models that we’ve built up, that we’ve proven to regulators are free of bias and the limitations of the model are well tracked, the stability of the models is well measured and monitored, I think that we can do great things in the world so I’m very excited about the future for Informed.
Peter: Wow, that is exciting so way beyond lending, it sounds like. Anyway, Justin, really appreciate your coming on the show today, thank you very much for spending time with us.
Justin: Thank you, Peter, it’s a delight to be here and thank you to Origence for giving us this opportunity to serve the thousand plus credit unions.
Peter: It’s really interesting to me that Informed have taken this technology to some of the largest banks in the country, they’ve taken it to the fintech lenders and now to the credit unions. Obviously, these types of companies are all quite different, but they all have the same needs and that is, you know, they want to be able to process documents in a quick and efficient way.
I’m excited about the credit union piece, in particular, because credit unions don’t have the technology capability, as I said, that some of the larger organizations have and now that we are able to have this really pretty sophisticated ability for when they’re processing loan applications to really make informed decisions quickly and that’s what it’s all about.
Anyway on that note, I will sign off. I very much appreciate you listening and I’ll catch you next time. Bye.
(music)
