By leveraging the benefits of artificial intelligence and looking at cash flow issues from the consumer’s perspective, Beem has grown to be one of the 50 most-downloaded financial apps in the world. It has been downloaded more than five million times in the two years since it emerged from stealth and has disbursed and recaptured more than $100 million without raising any debt.

It begins with CEO Akshay Krishnaiah’s experience growing up. He said his parents came from extreme poverty. One benefit of that was that they had to have complete control over their finances to survive. But even though his family was good with money, the world treated them differently, much worse than his well-off classmates who squandered many of their resources.

Why lower incomes have been ignored

Fast-forward to 2012, when Krishnaiah was working at PayPal’s product research division. That’s when he learned how the system worked and why people like him were treated differently. The insight? The risk profile of his group has never been properly analyzed and yet has often been mislabelled. Everyone was under one umbrella.

That’s a big parasol, like 100 million in the United States and more than three billion around the world big. It’s large enough that some have tried to cater to them, but they took the wrong approach. Krishnaiah explained that previous efforts were from a business engineering and product development standpoint. 

No matter how well-intentioned they were, they fell short. Because most leaders have never been underserved, they couldn’t understand the experience. Krishnaiah could. In case he didn’t, he took a year off and drove for Uber and Lyft in the San Francisco Bay Area to understand the plight of the gig worker and other passengers, 70% of whom were in the same boat.

Krishnaiah spoke with nurses, baristas, clerks and others who weren’t necessarily gig workers but shared their struggles. They may be working full-time. Their partner may be too. Yet it’s still a grind.

What Beem learned about the underserved

Before he could build a solution, Krishnaiah needed a compelling reason for folks to share their data so he could leverage artificial intelligence and machine learning. His year spent driving provided them.

“One of the things that I learned is that these people had cashflow challenges 10 times a year, and whenever they had them, it was often two to three times back to back,” Krishnaiah said. “If somebody or something could be delivered in a form factor that was easily understandable, usable and less intrusive, they would then be open to giving their data. Then we could apply AI, ML and other technologies to solve this problem and rectify the risk profile.”

Also read:

How Beem delivers results with less

A public benefit corporation, Beem has no credit checks, charges no interest and doesn’t have due dates. Give Krishnaiah a bank account, phone number, and date of birth, and he can deliver valuable insights that make clear differences in people’s lives. It’s based on the belief that the underserved are generally good people. The technology allows them to be more accurately assessed. Then it depends on what you do with it.

“The only differentiator as time progresses in the evolution of AI would be intent,” Krishnaiah explained. “I always tell my teams, partners and others that you can buy intellect, you can buy customers, and you can acquire customers, but you can never buy intent. It has to come from you, and that would become, as time progresses, the differentiator.”

Looking at people in a different way

Krishnaiah’s intent is to understand people holistically, eschewing traditional finance’s homogenous risk assessment methods. Those tools might credit the affluent for saving 20% of their income and penalize the barista for only saving $200. They punish women, whose healthcare expenses are 84% higher than men’s, while they make less and often have more caretaking responsibilities.

Why not instead assess how responsible someone is in their universe? How are they functioning long-term? Traditional methods don’t do that. They assume everyone shares the same DNA. Don’t adjust weights; remove income bias altogether. Rate people based on how they adjust to life’s seasonalities.

“We use technology to understand historically how they performed in their universe with all the moving parts and the data that they share with us,” Krishnaiah said. “Whether it’s personal information or information that we get from other sources for historical performance on the funding instruments, or other things that we can get about them from their transactions, all of this informs us, and then we try to understand how to create a financial responsibility index.”

Beem developed technology that predicts the real value of linked assets minus obligations, an often-ignored yet important factor. That allows them to predict the future and respectfully treat people within their universe. It recognizes that life is an index where conditions change periodically.

How it helps

Krishnaiah used the example of a widow with three daughters who relies on VA benefits, childcare benefits and other income sources. The sources are varied, and the amounts fluctuate, but she makes $30,000 a year. She had never before applied for credit and was now rejected because she had no score.

She didn’t have the co-pay for a dental bill. With Beem, she qualified for $70 within two minutes, which covered her copay. The remainder was put on a debit card she used to buy dinner for the family.

How Beem uses AI

Beem uses artificial intelligence in several ways. It begins with sanctions checks and negative news screening that forms links that could indicate fraud. Real-time transaction monitoring blends several AML and fraud typologies while expanding them with the help of multiple global partners.

Real-time risk rating provides different onboarding and user experiences for various risk profiles. They bring different verification levels. The process is augmented with human customer support co-piloted by AI.

Beem’s AI monitors transaction patterns to see if they match funding sources and practices. It also looks for previous accounts possibly related to or created by the same person. Potential fraud ring checks are performed by automating connected account discovery. It considers users within multiple hops who are strongly connected.

Data from multiple industry partners is deployed to perform hundreds of real-time checks before an account is established or service-level interactions have started. Little user input is required. A similar process is used for basic account data validation.

“We’ve been able to recycle $3 million to generate a throughput of $100 million. That has never happened before,” Krishnaiah said. “We have been able to do that because of the technologies.

“PayPal took 20 years to do what they did. Square took 10 to do what PayPal did. We’re doing it in five due to the pace of technological growth.”

Open Beem’s technology to all

Krishnaiah has big goals, ones he can’t accomplish alone. He wants to open Beem’s technology to others thanks to the AI Bricks philosophy that eliminates the multiple friction sources in a typical business funnel.

“But if you open up this technology, together we can all create an equitable society where you know the distribution of products, services and presentation of payments can happen equitably.

“That’s what powers Beem today – we call it AI Bricks. These are like Lego bricks that you can use individually or combine to get the same kind of success as we did.”