The fintech industry is highly competitive, and it’s important to have a moat around your business and/or your technology to carve your own niche. Upstart used proprietary machine learning and AI algorithms to develop its platform and ensure that it remains ahead of the pack. Though the AI technology disruption is at the nascent stage, Upstart founders believe it won’t be long when this technology will take over the entire lending ecosystem. The notion of using machine learning and artificial intelligence is to revamp how credit works fundamentally and not just as an assisting tech for online lending, fraud detection, and automation.
Impact of machine learning
The average credit card interest rate is 18.76 percent and $1,292 is paid by a household as credit card interest each year. Upstart, a fintech lender based out of San Carlos, California, claims to help save its borrowers almost 27% as compared to their credit card rates.
Future of machine learning
The young startup believes they are just starting out and that for machine learning to really kick in will require a ten-year investment into the process. Currently, the advantage is measured only in basis points, but after 10 years with the amount of data generated from 5 million originated loans, it will render the system unbeatable. The company has set a 4% loss rate, it could have easily focused on minimizing default rate but the goal is to expand and expand aggressively. The model has over 400 variables and usually considers on average 100 variables for an application. The company prides itself on developing everything in-house and it is no surprise the entire platform is built in-house, as well. So far, in terms of loan automation output, the company has managed to achieve 20% success, but it plans to achieve 80%-90% automation rate by the end of next year.
It has decided to sell its machine learning algorithms on a SaaS basis to banks, credit unions, and non-bank lenders. Its white label service allows point-of-sale financing for retailers. Its systems are instant and automated, thus facilitating any player looking to lend. The founders understand that their services can be replicated by a JP Morgan, but everyone does not enjoy the same scale or wishes to invest the millions and billions necessary for creating the prerequisite infrastructure. It services includes loan servicing rate request, credit modeling, and verification process as well. Pricing wise it has kept things pretty simple; there are no upfront fees. It is based on a pay as you use basis along with licensing fees.
The company is going a step further from the generic machine learning being executed today. It understands that human fraudsters are also utilizing machine learning to sharpen their skills. So the company is prioritizing adversarial machine learning versus machine learning in a static environment. The company believes it can never have a 100% foolproof system, but with enough data, it should reach a critical mass that creates statistically too many checkpoints for a fraudster to succeed on any meaningful scale.
Data is the new oil
Data is the new oil. And Upstart knows it. It is aware that the machine learning is something which can be replicated, but by pivoting to a SaaS model it is ensuring that it is able to extract data from multiple players versus being concentrated on its own in-house originations. This ensures that it will be the first to cross that Rubicon of critical mass when machine learning will become de rigueur for financing and Upstart will be the tallest if not the last man standing.
Upstart was founded in 2012 and initially started out as somewhat of a Kickstarter for human potential. It basically leveraged data around person’s educational background and other factors to help sell the person a predetermined percentage of their future cash flow for an initial lump sum payment. So using Upstart, X from Harvard Business School can sell 5% of his monthly income for 10 years for $100,000 today. Though the idea had a unique appeal, the company soon pivoted to the marketplace lending model.
Upstart offers innovative refinancing and lending solutions that help the borrower to consolidate their debts (student loans and credit card loans). Loans offered by the company are unsecured and terms available are between three to five years. Ever since its inception, the company has made rapid strides in terms of growth and it has done $640 million in loan origination and successfully originated over 50,000 loan applications. Its USP of offering lower APR is based on its ability to account for a person’s educational background and other qualitative factors to analyze their possibility of default and overall credit worthiness.
Funding to date
Initially, it raised $1.75 million seed capital from six investors and has managed to raise almost $85.0 million in various rounds of funding till date. The latest was a $32.5 million raised from four investors with Rakuten being the lead. With this cash injection, the company now sits on $40 million cash and is looking to achieve a cash flow positive status by the second half of the year. The company’s goal is not to become the biggest originator but a SaaS innovator focused on AI and machine learning.
Dave Girouard, Founder and CEO, has wealth of experience in information technology, his impressive CV includes positions at Google & Apple. Prior to this, he was working at Google as the President of Google Enterprise. Paul Gu, Co-Founder and Head of Product, previously worked in risk analysis at the D.E Shaw Group. Anna M. Counselman, Co-Founder and Head of Operations, was previously Head of Premium Services and Customer Programs at Google. The magnitude of the success and growth of the company can be measured by the fact that in a short span of time it has a team of over 200 employees.
Written by Heena Dhir. Edited by George Popescu.