Pecan.ai knows you値l take a mortgage before you do

Pecan.ai founders Noam Brezis and Zohar Bronfman Credit: Eyal Izhar
Pecan.ai founders Noam Brezis and Zohar Bronfman Credit: Eyal Izhar

After raising more than $100 million, the Israeli predictive analytics company's founders explain how their solution replaces data scientists and enables companies to compete with giants like Amazon and Google.

The formative moment when Pecan.ai founders Dr. Zohar Bronfman and Dr. Noam Brezis realized what they had in their hands came during an international competition of data scientists. Using data and AI processing, they had to predict which banking products it was worthwhile to offer customers of a certain bank. "We sat in a tiny room we’d rented at Tel Aviv University and realized there was a Miguel in Madrid who didn’t know yet that he was going to take out a mortgage, while we and our algorithm knew it before he did," Bronfman recalls. "I was in shock when I realized what had just happened. Awesome! We have a machine that forecasts what a person is going to do."

This happened four years ago, at one of the Kaggle website’s competitions for data scientists - "a kind of Olympics in which the winner becomes the master data scientist, like the victor in Israel’s annual International Bible Quiz," Bronfman jokes. "We had to do a lot of data preparation and this irritated Noam, who has this thing that if he gets annoyed, he takes a deep breath and attacks the problem. So he built, in one night, a data-preparation automization that became the prototype for Pecan."

Ultimately, the pair did not win the competition simply because they missed the deadline. But they decided, right before beginning their post-doctoral studies, to abandon academia and "do something that wasn’t detached, but instead was connected and made an impact." As Bronfman puts it, they didn’t just toss the prototype into a drawer. Instead, they transformed it into a platform for businesses that did not have the lavish funding and global glory of companies like Facebook and Amazon, but did understand they needed advanced analytical tools like AI in order to survive on the post-pandemic competitive battlefield.

To date, Pecan.ai has raised more than $115 million, of which $66 million arrived in February 2022 from investment funds including GV (formerly Google Ventures), GGV Capital, Dell Technologies Capital, Mindset Ventures and Vintage Investment Partners. The company has more than 100 employees in Israel and the US, and its prominent customers include healthcare company J&J, gaming giant Sciplay, Canada's CAA insurance company and Israel’s Phoenix insurance company.

Pecan.ai operates in a field called predictive analytics that has seen significant growth due to the disruptions and uncertainty in the business world in the wake of the Covid pandemic. Growing digitization, customer nervousness and intensifying competition, as well as the development of capabilities related to big data, AI and machine learning, have led businesses to understand that they are liable to lose their way without predictive analytics tools. According to the Research and Markets website, the field’s market share is expected to climb from $10.5 billion in 2021, to $28 billion within five years.

"Loyalty to a brand or a company has been undermined lately," Bronfman says. "Even customers who did only 5% of their shopping online raised it to 85% during the pandemic. So, there’s undoubtedly a rise in customer abandonment and a drop in loyalty, and companies are investing a lot more resources in customer retention. But this operates in a very narrow timeframe. If you’ve already decided to go elsewhere, the chance of retaining you as a customer is lower by tens of percentage points, than if you were still making up your mind."

In other words, the name of the game is early detection. Most companies try to identify trends by using data analytics that ask "human" questions, like who has reduced their number of visits to the site and who has bought less. By contrast, AI-based data analytics can create "questions" based on cross-tabulations that look completely arbitrary but will produce much more accurate rules, thus facilitating early forecasts regarding customers who are considering abandonment, as well as changes in consumer taste and consumption habits.

"The previous decade was characterized by business intelligence (BI)," Bronfman says. "All the organizations realized that they needed to collect data so they would know what’s going on. But in the past two years, companies have understood that BI isn’t enough anymore. Our customers report a 10% reduction in abandonment. In today’s world, shifting a major business variable at a double-digit rate is possible only by using AI."

Bronfman and Brezis crossed paths at Tel Aviv University when both of them were studying for a master’s degree in cognitive sciences, which deals with mapping and modeling neurological processes. Each of them arrived there from a different direction: Bronfman, 36, today lives in Tel Aviv, but grew up in Haifa. His parents immigrated to Israel from Ukraine in 1980. His father, public figure and immigration activist Dr. Roman Bronfman, is a media star these days as a commentator on Russia’s invasion of Ukraine. Although his parents came from Ukraine, Zohar Bronfman says he never felt like the child of immigrants: "My family got acclimatized at a very early stage - we spoke Hebrew at home." He served in the Israel Defense Forces’ elite intelligence unit, Unit 8200 and after his discharge studied philosophy and cognitive sciences.

Born in Jerusalem, Brezis, 41, lives in Hod Hasharon, north of Tel Aviv. He also served in Unit 8200. His late father, David Brezis, was a philosopher, and his mother, the economist Prof. Elise Brezis, specializes in macroeconomics. The family spoke French at home and the religious world played a major role in their lives. His mother came from a home affiliated with the Gerrer Hasidic ultra-Orthodox stream and his father took him to synagogue ("he always had a philosophy book tucked inside his prayerbook," Brezis recalls). From an early age, Brezis had what he calls "a passion for computers."

"I felt like I had no understanding whatsoever of what they were talking about at home, they had high-level philosophical conversations. So I went into my own world. As a boy, I had textbooks about software languages and I fell so much in love with them that I read them over and over again."

From 2009 to 2016, after his discharge from military service, Brezis consulted for companies on business intelligence and lectured on the topic. At the same time, he earned a bachelor’s degree in economics and psychology because "in our home, it was unthinkable that you wouldn’t go and study economics. Because of my mother."

Brezis met Bronfman, while both were studying for a master’s degree in cognitive sciences, and their lives intertwined. "On the first day of studies, I spotted Noam and immediately sat next to him, and we started to talk," Bronfman says. "We connected within about 20 minutes. Since then, we’ve done everything together, every research study, every article and every publication."

The two men continued at Tel Aviv University and earned their Ph.Ds., which dealt with modeling for cognition and neuroscience. Afterward, they began planning an idea and a place for post-doctoral studies and deciding where they would move with the families they had established in the meantime.

"We went to the gym together lots of times and the trainer would laugh at us: ‘Why do you bother coming here,’ he’d say, ‘you don’t even work out - all you do is talk,’" Brezis recalls. "The post-doctorate was a natural continuation for us, but Zohar dreamed of doing something with AI. So I said to him if that’s your dream, why do we have to do this whole post-doc thing? Let’s take our ideas and do something real with them, something that can be used."

Bronfman adds: "I already saw myself riding up in the elevator of the ivory tower and sitting very comfortably in academia. But for Noam, it was important that we do something that wasn’t detached, something connected that would make an impact - this was more important to him than my desire for academic life was to me. What was important to me was to be with Noam. People here find it funny that my very first payslip came from Pecan."

Not only in the dazzling companies

To explain what Pecan.ai does, it’s worth understanding first what data science is. It’s a relatively new field that uses large databases (big data) and puts them through complex statistical processing in order to obtain accurate forecasts. For example, take one of businesses’ most painful problems - customer abandonment. Few companies can afford to access the services of data scientists, who have deep and specific qualifications: a master’s or a doctorate specializing in statistics, data-processing techniques and so on. "Data scientists are a scarce resource," Bronfman notes, "and they always go to the dazzling companies."

"One of our customers, for example, is a US company that deals with Botox and beauty, and they want to be precise about calls to their call center. They want to know who is calling and what to offer them," Brezis says. "A company like this doesn’t have a chance of recruiting a data scientist, who will usually prefer to go to Facebook, Google, Amazon or to something ethical, like healthcare technology."

This is where Pecan comes into the story: The company, in essence, offers access to data science and provides a platform that replaces the human data analyst, similar to the way in which Wix saves its users from having to hire a website builder. This, however, applies only partially: Using Pecan’s platform does require a certain amount of training as a data analyst who can do the calculations that fall under the heading "business intelligence," but such analysts are actually the most commonly found.

"Our biggest source of pride is to see how these analysts add ‘data scientist’ to their LinkedIn profiles after they work with our platform," Bronfman says.

In other words, you are democratizing AI and transforming it from a tool used by Amazon and Facebook into a tool that’s accessible to a wide range of companies.

Brezis: "We’re making it accessible to large groups that haven’t yet managed to get past the hurdle of moving from BI analyses to using data science and AI."

The major advantage of AI is the ability to spot at an earlier stage the risk that a customer might leave, when they are still hesitating about it; thus, the chances of getting them to stay are much higher. How does AI do this? It crisscrosses things that arise from human logic, such as change (even minuscule change) in the customer’s consumption behavior, with all kinds of variables that would not arise from the human brain. This combination produces rules that look absolutely arbitrary to the brain - but they work. "This is logic we don’t know how to imitate," Bronfman observes. "The advantage is that these rules are much more precise than human rules. The disadvantage is that you don’t always understand the logic."

AI can also make decisions without even knowing they are not ethical, and you won’t know it either. For example, people with certain psychological problems will tend to buy much more if you bombard them with deals.

"Absolutely, and anyone who uses AI has to take this into account."

And do you take this into account, or do you just give the data to the customer, who will or will not apply ethical restrictions?

"We produce technology and build our tool like you build a computer. It can be used to fire a nuclear missile and it can be used to develop medications. But we’re based on data that doesn’t contain improper variables. There are no private, identifiable variables - and of course, there aren’t any medical or demographic variables."

People talk about a severe shortage of personnel in tech. On the other hand, Wix comes along and makes website builders redundant, while you’re bypassing data scientists. Will a lot of techies ultimately become redundant?

Bronfman: "It’s happening. Lots of startups are trying to take things that people do and produce automated versions. But at least in our field of data science, it’s clear that people aren’t redundant - instead, they turn to dealing with problems that can’t be automated. And there are a lot of those."

Instigating the AI revolution

Back to the birth of Pecan.ai. At the beginning of 2018, two months after finishing their doctorates, Bronfman and Brezis rented a small room at Tel Aviv University and started to work. "We were two people with very limited business backgrounds, but we started to speak with people, to tell them about our idea of making automated data science accessible, and we met with several venture capital funds," Bronfman says.

Among the funds they met with, the two entrepreneurs were captivated by Haim Sadger and Aya Peterburg, the founders of S Capital, whose $4 million investment in Pecan started the company on its path. At the time, Brezis and Bronfman had the prototype they created for the competition they had participated in, as well as a few pilot customers they got from anywhere and everywhere, or, as Brezis puts it, "all kinds of friends and people we met. It was all very piratical."

The initial stage was building the first version of the product. "We started to market it and we saw that companies had an almost existential need for these capabilities. We held discussions with very senior people and very large organizations, and the deals moved very quickly. The companies realized that this was already an existential need, that they had to pull rabbits of the hat to beat the competition.

"Everyone is arming themselves with data now and everyone is building digital capabilities, so if you don’t leverage your data, you simply won’t survive in the new business environment," Bronfman points out.

Why is it precisely now that companies are realizing they need to pull rabbits out of the hat?

Bronfman: "It’s a matter of educating the market. It takes time for processes to ripen. The previous decade was characterized by BI. But today, when you look at the secret of success for leading ecommerce companies like Amazon, you realize they have very smart algorithms that leverage their data so they can do personalization and optimize their pricing and supply chains. These are things you could never achieve just by using business intelligence. The moment you understand that the industry’s leading lights use AI, you start seeking ways to level the playing field. This enabled us to triple our sales twice, once from 2019 to 2020, and the second time from 2020 to 2021."

"We’re just scratching the tip of the iceberg. The gap between supply and demand in AI and data science capabilities is tremendous. We always laugh about the fact that in our investment presentation, we never had a slide about market size. It’s like saying how much water is in the ocean. We want to bring about the revolution of AI as a basic tool that every business team is able to use. So we’re trying to build a very large company."

Dr. Noam Brezis, 41 years old

Personal: Married with three children, lives in Hod Hasharon north of Tel Aviv

Cofounder and CTO at Pecan.ai

Dr. Zohar Bronfman, 36 years old

Personal: Married with two children, lives in the Tel Aviv suburb of Ramat Aviv

Cofounder and CEO at Pecan.ai

Published by Globes, Israel business news - en.globes.co.il - on July 10, 2022.

© Copyright of Globes Publisher Itonut (1983) Ltd., 2022.

Pecan.ai founders Noam Brezis and Zohar Bronfman Credit: Eyal Izhar
Pecan.ai founders Noam Brezis and Zohar Bronfman Credit: Eyal Izhar
Twitter Facebook Linkedin RSS Newsletters 粮裔 Israel Business Conference 2018