UPST has a real competitor, Zest AI?

I am seeking anyone’s input on the topic of Upstart versus Zest AI.
While digging around to search for other potential competitors to Upstart I’ve been reading up on a privately held company called Zest AI. Previously known as Zest Finance, the company was founded in May 2009 by Douglas Merrill, former CIO and VP of Engineering at Google.
Mr. Merrill founded ZestFinance with a hypothesis: “Google-like algorithms could be applied to make consumer credit more transparent, available to more people, and significantly less expensive. ZestFinance’s team of crazy smart data scientists and mathematicians are united by a unique mission: to enable fair and transparent credit for everyone."

This sounds quite familiar to us, an almost identical story to Upstart. Both were founded by ex Googlers. Both with a mission to use AI/ML for alternative and better methods of credit underwriting.

Zest AI had a 6 year headstart in this space from 2009, as Upstart began its mission in consumer lending in 2015.

And yet, Upstart is the company currently valued at $11 billion, and profitable, with 600M revenues expected for FY21…while Zest AI appears to have a valuation far below $1 billion (about $90M it seems?, not sure how reliable as the only source I can find is here: https://dot.la/startups-in-los-angeles-2649633594.html) and revenue estimated below $100M.

Founder Douglas Merrill no longer leads the company as of Nov 2019. Zest’s cofounder Sonya Merrill, (wife of Douglas Merrill) was Zest’s chief people officer and also Ex Googler, former head of communications (emerging markets). She left ZestFinance in Dec 2019.
Kasia Chmielinski had joined the team June 2010 as VP Product management. Also an ex-googler, former communications manager. But Kasia left ZestFinance in June 2014.

It seems a case of mis-management by leadership and misguided execution for nearly 11 years as to the reason why the company is valued at less than Upstart (for now?).

But, I do think they are now a very valid competitor against Upstart since the old management team has left.

So what happened? Let’s follow the timeline.

2010: ZestCash introduced. This was the original name and focus of the company. It was initially founded to provide “cheaper’ payday loans to struggling consumers via alternative credit underwriting.

April 2012: ZestCash introduced Hollerith, a “new set of underwriting models that allow the company to extend credit to 25 percent more Americans and increase repayment from customers by 20 percent.” They claim “ZestCash used to run one big data underwriting model that looked at up to a thousand variables”. ZestCash now runs about 10 unique underwriting models simultaneously that then transform this data into tens of thousands of useful meta variables. https://techcrunch.com/2012/04/26/zestcash-debuts-new-big-da…

However, it seems the company eventually abandoned this ZestCash venture as the about page no longer includes anything from Zest: https://www.zestcash.com/about

July 2013: ZestCash renamed as ZestFinance; raised 20M from Peter Thiel and others. At that time, since the company’s inception, ZestFinance had “increased net repayment by 90 percent over industry scores and more than doubled the number of underbanked Americans the company serves.”

October 2014: “After rejecting two-thirds of applicants, ZestFinance approves loans that average $600 for those that make the cut. The borrower pays $400 in interest for a six-month $600 loan. That computes to an annual percentage rate, or APR, of 390 percent. Merrill says he only breaks even on the loans, since he spends far more than prime lenders do for capital, marketing and the write-off of defaulted debt.”
“Since 2009, the company has made 100,000 loans through a Web site called ZestCash and another called Spotloan. [this will lead ZestFinance into trouble later] Spotloan is owned by the Turtle Mountain band of the Chippewa Indian tribe of North Dakota, which asserts it isn’t subject to state laws.” https://www.washingtonpost.com/business/zestfinance-issues-s…

June 2015: JD.com and ZestFinance had a joint venture announced to “leverage ZestFinance’s machine learning-based credit-decisioning technology and JD.com’s reservoir of consumer data to provide credit risk evaluation services”.

July 2015: ZestFinance decided to move past focusing on sub-prime borrowers, and onto “almost-prime or near-prime” borrowers with the release of Basix– unsecured personal loans – which will be available for a three-year term at fixed annual percentage rates (APR) from 26% to 36%. Monthly payment amounts and APR do not change over the course of the loan. However, it seems the company eventually abandoned this just like with ZestCash, despite raising $150M in funds from Fortress Investment Group in that year for Basix. https://www.crowdfundinsider.com/2015/10/75391-three-month-o…

Basix’s website today: https://basixloan.net/about-us.html which appears to show nothing to do with Zest AI.

July 2016: ZestFinance received an investment of $30 million in 2016 from Baidu, giving the company a $252 million valuation, according to Pitchbook. China’s fast growing credit market lacks a centralized credit scoring system.
February 2017: ZestFinance introduced the Zest Automated Machine Learning (ZAML) platform, opening up the technology to any bank, credit card issuer or auto finance company. Here we see the company pivot further to a B2B strategy instead of direct to consumer lending.
August 2017: ZestFinance and Ford’s Motor Credit company team up after their machine learning study compared results from a Ford Credit scoring model with a machine learning model developed by ZestFinance. Ford Credit and ZestFinance found that machine learning-based underwriting could reduce future credit losses significantly and potentially improve approval rates for qualified consumers, while maintaining its consistent underwriting standards.
October 2018: Class action lawsuit filed against ZestFinance. It alleges that the defendants are part of a “rent-a-tribe scheme” for making payday loans. Non-tribal entities ZestFinance and Merrill allegedly provided the capital, marketing, underwriting and other resources for BlueChip Financial, doing business as Spotloan. The suit alleges that borrowers were charged rates above what is allowed in Washington – as high as 490% annual interest. https://jnswire.s3.amazonaws.com/jns-media/da/6f/1164563/ZES…
https://legalnewsline.com/stories/512366003-former-google-ex…

December 2018: ZestFinance and Microsoft partners up. “Financial institutions will be able to use the Zest Automated Machine Learning (ZAML) tools to build, deploy and monitor credit models using the Microsoft Azure cloud and Machine Learning Server. A customer of both Microsoft and Zest, Prestige Financial Services, implemented the explainable ML credit scoring solution in early 2018. Within six months, Prestige doubled its lending volume and made credit to consumers previously overlooked by legacy models.”

March 2019: ZestFinance entered an agreement with Discover to create “the world’s biggest platform for AI-based credit scoring”. Discover will use the Zest automated machine learning platform for a more accurate lending criteria beyond the use of traditional credit scores.

March 2019: ZestFinance launches ZAML Fair. “Several mortgage lenders have already tested the ZAML Fair algorithm. Based on those results, ZAML Fair would eliminate 70% of the gap in U.S. mortgage approval rates between Hispanic and white borrowers and cut the even larger gap between black and white borrowers by more than 40%.”

July 2019: ZestFinance announces partnership with Brazilian retail company Via Varejo to use Zest ML to deliver a more precise credit decision immediately, safely expanding its revenue and reducing the risk of bad loans for point of sale microfinancing.

October 2019: ZestFinance announced integration with MeridianLink, which provides loan orignation software and has 1,925 banks, credit unions, mortgage lenders, specialty lending providers, and CRAs. (MeridianLink filed S1 on April 30 2021). MeridianLink will integrate with ZAML credit scoring directly to its LoansPQ platform, providing clients the ability to access advanced machine learning for lending.

October 2019: ZestFinance founder Douglas Merrill, is no longer the CEO and appears to have left the company in this month.

October 2019: ZestFinance rebrands to Zest AI. “In the two years since Zest AI introduced its flagship product Zest Automated Machine Learning (ZAML), the company has achieved annual triple-digit percentage growth. Zest AI powers $500 billion in total lending by clients in every geography and credit category, including credit cards, personal loans, auto, mortgages, and student loans. With a quarter-million applicants scored every month, Zest software has helped 10 million more consumers get access to credit. ZAML clients typically achieve a 15 percent increase in approval rates with no added risk or a 30 percent reduction in losses keeping approval rates constant. Clients to date have generated $1 billion in profit from reduced credit losses. Lenders can also more easily approve borrowers with limited credit history, in some cases by up to 5x. With the use of Zest’s bias-reduction tools, lenders can shrink the gap in approval rates between white and minority borrowers by 30 percent.”

October 2019: VyStar is the first credit union to deploy ZAML for underwriting. Vystar (13th largest credit union and $11 billion in assets) will deploy Zest machine learning in member credit card services first, followed by auto loans in early 2020

November 2019: defi Solutions, an innovative lending technology and BPO services provider, partners with Zest AI to use their ZAML for loan decisioning.

February 2020: ZestFinance agreed to pay $18M and cancel $170M of outstanding debt to end proposed class litigation alleging they saddled consumers with payday loans carrying exorbitant triple digit interest rates. It looks like they exited completely from direct to consumer lending by this point?
https://buckleyfirm.com/sites/default/files/Buckley%20InfoBy…
Titus v. ZestFinance, Inc., et al. USDC W.D. Wa. 3:18-cv-05373
Bulette v. ZestFinance, Inc., et al. Cal. Sup. Ct. Alameda Cnty. HG18929579
Turner, et al., v. ZestFinance, Inc., et al. USDC E.D. Va. 3: l 9-cv-00293

March 2020: Zest AI laid off roughly 20% of its 100-person staff when Covid hit in March, according to LinkedIn data.

October 2020: Zest AI closes 15M funding round from Insight Partners. Zest AI at the time claimed that on average, its clients see a 20% increase in approval rates with no added risk and 30% reductions in charge-offs. Clients can integrate its technology to build their own models and meet compliance with various regulations. “Its pre-money valuation was at $75 million, according to Pitchbook.” (https://dot.la/startups-in-los-angeles-2649633594/playvs) CEO De Vere said when he shows would-be clients that his company’s ML software package offers a 10- to 50-times return on investment, “it becomes a very easy pill to swallow.” De Vere said that when he sits down with large banks that might have 1,000 data scientists, “we’re all talking to each other and everyone knows that machine learning is where it’s at. https://www.pymnts.com/loans/2020/zest-ai-ceo-artificial-int…

They also have partnerships with Turkey’s third largest bank, Akbank, which “allowed it to reduce its nonperforming loans by 45%, and double its approval rate for borrowers with no credit history.”
Zest AI also has work with BNP Paribas in France.

November 2020: Freddie Mac, the company created by the U.S. government in 1970 to fund mortgages and increase home ownership, will start using fintech company Zest AI to help assess the risk of people defaulting on their mortgages.

June 2021: Zest AI announced an $18 million capital injection led by strategic investors VyStar Credit Union and First National Bank of Omaha

So to summarize.
12 years ago, Zest AI was created as a payday direct to consumer lender in the subprime market. They were among the first, way before Upstart, that I know of, to utilize machine learning and truly alternative data for credit underwriting models. That venture was not profitable and got them into huge trouble with a class action lawsuit. Around 4 years ago they pivoted to a SaaS business, selling their AI/ML underwriting to financial institutions. They claim to have improved approval rates and loss rates, similar to Upstart’s results. We can also see they have lots of patents filed for their models https://patents.justia.com/inventor/douglas-c-merrill

Since old management was kicked out two years ago, they have aggressively expanded their footprint further with many different leading institutions going into 2021, which includes names such as Discover and Ford, and even with companies in international/outside geographies like Brazil, France, China or Turkey.
Although they don’t have a No Action Letter from CFPB like Upstart, they appear to be doing well on the compliance front, as they even got Freddie Mac to sign on.

Key differences from Upstart:

  1. They are selling only their underwriting abilities, not as a White label service like Upstart is doing, but by helping each institution ‘build’ their own models.
    Therefore, they don’t appear to have much influence on the consumer experience, like Upstart does with their direct website/white label app, and Upstart’s expansion into the auto dealership front with Prodigy. So, it seems ZestAI won’t have much control on customer satisfaction which otherwise could help drive loan volumes for banks.

  2. They have their hands on everything as much as possible: different international geographies and the entire spectrum of credit. They didn’t start out focused entirely on personal loans before moving on; they’re in credit cards, auto, mortgages, point of sale microfinancing, etc.
    Their company size is still quite small, at 88 employees. (Upstart has 800+) It’s almost hard to believe as they have so many partnerships with big institutions and internationally, and still be focused on working on all the different types of credit.

  3. Management is not up to par as Upstart’s, but maybe improving?
    Limited reviews on glassdoor (wow, only 33% approve of current CEO) but some examples: https://www.glassdoor.com/Reviews/Zest-AI-Reviews-E610161.ht…
    Sep 2020: Most existing organizational issues have seen significant progress over last year with leadership effectively communicating changes
    Nov 2020: Total Toxic workplace dominated by unreliable, and unavailable senior Sales and Exec team. This company feels like a used car dealership that somehow stumbled across AI/ML and stitched together a slick sounding software.
    Aug 2020: Company changed from a culture that was about doing the best work to one that is about selling whatever a customer will buy.
    May 2020: Recent management change. New management is far less competent and has been cutting way back on benefits. Hiring standards have gone way down. Big customers won’t save you.

  4. Culture is not the same as Upstart. They don’t talk much about trying to lower interest rates to consumers, their website is all about “here’s how we can help your instiution increase approval rates and avoid losses”. And they had to pay up big in their class action lawsuit, after overcharging subprime folks with payday loans.
    2014 Glassdoor review: “The interest rates may be lower than payday loans, but not by much, so borrowers are paying a lot of interest which may raise some ethical questions.”

Despite all their past troubles, on the surface I really do think they are poised to be a strong competitor, even with the massive TAMs in consumer lending. They seem to have the tech and access to a lot of data and have big name customers going for them. If they execute quickly and successfully they can get a good network effect going for them.
Their 2019 press release claimed they were helping to underwrite 250,000 customers per month and already $500 billion in total lending. I just don’t get why their private valuation is so tiny, which was $90 million as of last year (Unless someone here has better/accurate info?)

I’m interested in everyone’s thoughts.

55 Likes

Been lurking here for about 3 months and this is my first post. I’d like to thank everyone here that contributes; the brain power here is off the charts. Hopefully this is a worthwhile add.

I found this 47 minute video between Zest AI CEO, Mike de Vere, and Jenny Vipperman, CLO of VyStar Credit Union (Zest’s biggest CU client): https://vimeo.com/570004452

I’ve been an investor in private equity for the past 10 years. Over that time, I’ve viewed 100’s of presentations from private company CEOs. My impression of Mr. de Vere is that he falls in the bottom 50%. Compared to the videos I’ve watched from the Upstart execs, he offers a stark contrast.

Ms. Vipperman on the other hand is outstanding. Upstart should hire her to sell to credit unions. She is the poster child for CUs to adopt AI. Here are some notes from some of the things she said:

*AI means you suspend everything you know about lending.
*AI is about fairness.
*Regulators want better decisions.
-Is the AI transparent and fair?
-We use outside sources to test results.
-We believe outcomes are far better using AI. Not said, but implied → Regulators will love this.
*VyStar went from 40% instant approval to 60% after implementing AI. This allows us to grow without growing headcount. Our loan officers have more time to look at other stuff (that AI has yet to provide a solution for).
*Days of using generalized credit scores are over.
*Different products use different AI models; allows us to offer risker loans for good of the community.
*AI allows us to improve our service to members.
*AI will allow us to be on par with the big banks at which point we win based on our heart and soul.
*Destination is breaking the process (the stress members go through to get loans); AI is the vehicle to get there (to a faster, more streamlined approval process).
*Not only do we use AI to approve credit card applications (faster and a higher percentage), our average credit limits have increased by 50%.
*We have to move down this path.
*AI is a gamechanger transforming the way we do lending.

Upstart’s endorsement from the CU association should give them a huge edge over Zest. When the decision goes from should we use AI to which AI vendor do we go with, adoption goes from first gear into overdrive. So yes, Zest is a competitor but I think is a very good thing. The word is getting out as you see from the video; it was done for an audience of CUs.

My chips are on UPST over Zest AI. If Zest becomes a real competitor, UPST buys them, end of story. More likely now as they are no longer founder led but rather VC controlled.

–FooLforBCS
Long UPST, 4.4%. When I first came to this board, I owned over 50 stocks (and no UPST). In three months, I’ve pared that down to 25. Not sure if I will be able to get 10 but maybe someday. I am now 79.9% Saul stocks (defined as owned by at least 2 members on the google sheet), 9.6% cash and 10.5% legacy positions that are on the chopping/evaluation block.

92 Likes