Catherine Flax interview

Catherine Flax is the CEO of Pefin

Q: Welcome, Catherine. Tell us about how you got started in the industry.

A: Working in large financial institutions with innovation boards and innovation labs made me think about what is happening in banking, what the role of banks should be, how that will change, and what parts of banking will get disintermediated by technology. Having things like algorithmic trading as part of my business gave me a glimpse into what will change. Voice brokers used to be how we did business, but now they’re all gone. Everything is on the screen. So, for me, that began my journey into learning how technology will change our businesses and it made me want to understand that better.

Being on innovation boards, I was able to see all the Fin-Tech companies that would come to the bank and pitch a lot of different areas in machine-learning and robotics – Big-data, Blockchain, and all those different things. They would try to piece together what is happening now, what is useful, and what aspect of the business it will help us with. Is it cost reduction? Is it efficiency increasing? Is it better meeting the customer’s needs? There are a lot of different things that were going on. Through that, the bank made an investment in a Blockchain company. They asked me to sit on the board. This was interesting both in terms of being able to dive deeper into what Blockchain is and why it matters, but also being able to watch from the beginning of a start-up how does that all work, and what is different in that work environment compared to what we are doing at the bank. I also learned how to better make decisions, execute, learn, and change direction. This becomes very difficult in a large bureaucratic organization.

Q: Of course.

A: At the same time, a good friend of mine started a company, Pefin. Her background is also in machine-learning and financial engineering, and she was in the algorithmic trading group at Goldman Sachs, when her dad lost his job. He was in his early sixties, and he wasn’t in finance – he was an engineer. He was really panicked about whether he needed to look for another job, if he could afford early retirement, and what he would do. So, she spent a lot of time with her dad helping him to pull together and understand his financial situation. But she was also digging into things like how social security works, what Medicare is, and other things that neither of them knew. Together, they did all this research and she was able to show him that it was okay if he retired. That was a huge thing for him, but for her it was also a real watershed moment. She was thinking “There are a lot of smart people out there like my dad. They have no idea how any of this works and they are never going to see a financial advisor. They might not have enough money, or they might not trust that they will receive the right answer.”

Eventually, she quit her job at Goldman and set out to build a company. She wasn’t thinking, “I know about AI, let me find the problem.” She just saw a problem, and she wanted to fix it. Because of the nature of the problem, the fact that it is data-intensive, and the fact that the relationships among the data mean that if it were approached from a rule-based perspective it would be intractable, she decided AI was the right answer to solve the problem. Fortunately, she had a background in that, so she started on that path.

Q: Great!

A: Because she was a friend, we were always talking about this. She shared with me what she was doing, and about two years ago she asked me to become a more formal advisor to the company for a couple of reasons. First, because they were in beta at the time, so they were receiving a lot of reverse inquiry from several large institutions globally asking, “Is there some way we can partner?” “How does this work?” Her background is not in large institutional business development, so she came to the conclusion that she needed a CEO. She asked me to help her conduct a CEO search. I started as an advisor that just helped them occasionally. I never could have imagined that we would end here. But last fall she came to me and asked, “Would you be interested in doing this?” It was an interesting process for me to consider that.

Q: Where were you at the time?

A: At that point, I was at the BNP Paribas running Commodities, Foreign Exchange, and Emerging Markets, sales and trading.

Q: So, something different?

A: It was a very different career-path decision, but because of all the thinking I had been doing about technological change it wasn’t so crazy. Although, I knew that it would be a big change in terms of just what it meant to go to work every day because it was completely different to work in a start-up. Fortunately, I have a number of friends who are CEO’s at technology companies. Some of them came out of banking, and some did not. I spent a lot of time with all of them trying to understand what I did not know about this, and what I was missing.

One of them, a former banker who runs a technology company out of Germany, said to me “You have to know that most technology companies fail. If that is not going to ruin your life than it is a really good thing to know.” Going into this I was like, “It’s cool. I can do that.” Then he said “You have no idea how spoiled you’ve gotten working in banking. You have assistants, business class flights, and all that stuff.” At first, I thought I didn’t really care about that stuff, so it was fine. However, I realized when I got there that he did not really mean that I was spoiled because of those perks, what he meant was when you are at a company like JPMorgan or BNP Paribas you have so many resources at your disposal.

Q: That make sense.

A: So, as you know, I was mostly in sales and trading. At one point I was the Chief Marketing Officer for JPMorgan where there were 700 people in the marketing division globally. If you come up with an idea, somewhere in that 700 people there is a sub-set that is an army who you can get to sort it out. It is not that you’re not working hard, it is just that you are generally operating at a high-level. The days that you were really like “in the weeds” were twenty-five years ago when you were building spreadsheets, and you were doing your junior jobs. What I found being in a start-up is that you are constantly flexing between the most minute detail and the big issues of “Where are we going with it?” It can be challenging but it is also part of what is really interesting and fun about the experience. I think, for me, that was probably the biggest change.

Q: So, the flexing between…

A: I think that is the biggest change. Obviously, as the company grows you will have more people who do that more tactical stuff. But, for me, it has probably been one of the best growth opportunities in my career because it is the only way to really know what is going on.

Q:  What makes you say that?

A: Typically, for somebody at this stage in their career, I would have just continued with the high-level if I had stayed in banking.

Q: Of course.

A: I don’t think the world works like that anymore. As I mentioned earlier, this is something that I was speaking to a group of bankers about yesterday The conversation came up around buying versus building new platforms, and how you innovate within a bank. One of the things I said to them was, “If you’re prepared to change the way that you do your development, receive information from your customers, and you are willing to adjust your platform – then congratulations, you could be ready to build this yourself. If you aren’t ready, then you may want to partner with a company that is.” People don’t have a lot of patience anymore when someone tells them there’s a problem, without fixing it. I think that’s a hard thing for large bureaucratic institutions to internalize. They don’t understand how to make those adaptations to the fundamental business model.

Q: Can you tell us a little bit about how the AI works in your company?

A: As you know, I believe when a lot of people think AI they either think about Amazon, which gives you shopping suggestions, or Facebook, which can identify a picture of a cat.

Q: Right.

A: It works great if you want to identify a cat. However, it doesn’t work as well if you’re looking for something very specific to you, such as medicinal or personal financial advice. Ultimately, somebody “like you” isn’t you. Getting personalized advice could be really bad, so it’s not a good idea.

So, that is generalized AI. Although we do use that for things like content, as the platform does have a lot of content creation. It is possible that if you told the platform you want to have a baby and buy a house, then suggesting an article is actually a good way of using generalized AI – which we do. But more importantly, how do you understand people in the context of their life and who they are?

There are two other kinds of AI that we use in the platform. One is a neural network, so it understands what those relationships are for that individual across all the different dimensions of their life. Then, once it understands those relationships, it figures out how you propagate that forward until the expectation of their death. If somebody says, “I live in New York City and here are all of the dimensions of my life” there are hundreds, which become millions, of data points as you go throughout their life. If they say, “But I’d like to see what it looks like if I move to Connecticut” – you also need to construct it in such a way, that they do not just get housing cost change, but also the tax change and the day-care cost change. Many things in their life are going to change, and it all has to update instantaneously as everything goes forward. A lot of what happened at Pefin from an innovation perspective has been about how to process that much data and make it both web and mobile friendly, so it can work for people.

The other aspect of AI is reinforcement learning. Really understanding what the spending behaviors are and the interactions that people have with their money. When people come onto the platform they connect their accounts by using aggregation -your credit cards and banking accounts and all of that -so typically the platform gets three months of information, which is not enough. So, initially it is going to make some assumptions. However, over the course of time the platform is going to learn better if you are a spender or a saver, how that looks, and what the seasonality is in your spending behaviors. This really improves the predictive abilities of the platform, but it also allows it to give people more insights about what they are doing. Typically, people do not have the time or the bandwidth to be that introspective about their behaviors. Being able to incorporate that into the platform helps people to stay on track with things that they have said they wanted to do. That is what Pefin does.

Q: That is really cool! So, in these two areas (neural nets and reinforcement learning), has it been hard to acquire talent? It must be very competitive.

A: It is very competitive. However, we are really fortunate because the people in this space are really excited about the topic!

Q: That’s good!

A: We are hiring right now, and we have been flooded with resumes. People are really into it. That is the good news, but the bad news is this is a very competitive space. A lot of other people are trying to hire as well. The battle for talent is ongoing, and part of the problem talent is location – it can be anywhere in the world.

Q: Yeah.

A: Now, we have a partnership with a client in the Netherlands – we are building out a platform for a large pension fund. We are now in the process of establishing an office there, which I think from a geographical diversity perspective is going to help us in terms of Intel and talent acquisition. Right now, everyone we have with the exception of one person, is in New York City.

Q: Can you tell us a little bit about the business model?

A: We have the direct-to-consumer and the B2B model. The direct-to-consumer platform launched at the end of 2017, so we are only weeks into that. We did a soft launch, partly because there were a couple of important features that we didn’t have in beta. So we thought, “Well, let’s go ahead and launch because we can’t not launch forever. We should just do it, then we can see how it goes until we need to make some modifications.” One big change from beta to the launch was that in beta everything was about planning and advice. If it was appropriate for someone to have a taxable investment account, we would just say, “Here is the allocation that we would recommend for you. Do with that what you will.” We had a lot of beta users come to us and say, “Can’t you just execute that for me?”. So, we built a registered investment advisor as a subsidiary, and we launched with that at the end of the year so that people could execute their trades. We still tell them they don’t have to execute through us, as philosophically it is important to us that they know we are just giving them the advice we think is right. They do not need to invest through us.

The other difference between the launch and beta is that, initially, in beta when people first came onto the platform they would input a lot of information. There is a timeline that is the summary of all of their information, so they can do a lot of scenario analysis and playing around. They can ask questions like, “What if I wanted to have a baby?”, or something like that. That is the most powerful part of the platform. Really allowing people to see things like if they downsize that home, or if they move their retirement out a couple of years, what that all could mean for them. But what we found in beta was that some users just jumped right into that and said, “Let me play around with that” some were kind of intimidated by “How do I do that?” So, we started having a guided chat process. It is very deterministic as the chat goes. But it will help users say, “Okay, would you like to plan for your child to go to college?” “Yes” “Do you know what college you want to send them to?” “Yes”. Instead of them having to figure that out, it makes suggestions, like “Your plans are only 82% achievable, would you consider moving your retirement out by two years, or would you consider reducing your discretionary spending by 5%?” It helps them understand their options, then once they get into it then they can go and do their own scenario analysis.

What we have found since we launched is that the deterministic chat is still a bit too long. So, we are making changes right now to the on-boarding because it feels like we need to give people smaller, bite-sized pieces of information. We do not want to make them feel lost. That is the part of the soft launch. Long story short, we are going to be doing more PR and marketing later in the year. As I mentioned earlier, we are going through a capital raise, which is a bit of what we are waiting for. Once all of that is done and we make some of those tweaks to the platform, then we are going to do more full-court press on the direct-consumer business.

On the B2B side it has been really interesting. On the one hand, it’s great that we are building the platform out with the folks in the Netherlands, partly because its revenue, which is always helpful. But also, it is a great learning process to see what is different about building this platform out in a different culture, in a different country, and in a place where some of the issues are the same but a lot of are not. Certainly, medical and college costs are different- these aren’t the burden in the Netherlands that they are in the US. It has been a really great education for us to do that.

Basically, there are three general buckets of B2B clients that we are speaking with right now. One is to “white label” the product for financial institutions.

The second one is in employee benefits. There is a huge surge right now in companies understanding that financial wellness is a real issue for their employees. How do they incorporate that into their overall wellness programs?  We have had a lot of companies reach out to us about distributing the platform to their employees, which can take a lot of different shapes. It either requires a lot of customization or very little customization depending on how they want to implement, so there is a lot of conversation going on there, and even that has been interesting. I had somebody reach out to me a few weeks ago. That individual gets hired by big companies to come in and do seminars on personal finance. The guy called me and said, “I love what I do, but I am also very cognizant of the fact that a one-hour seminar is not really going to do it for people. It might raise awareness that there is something they could do, but then I leave them with nothing- no tools to actually make a change.” He was suggesting that perhaps we can partner and have Pefin as the tool for them, to be able to say, “You know you need to do something, here is how you do it.”

Then the third bucket of B2B business is around the financial advisor. Again, it is taking a lot of different shapes where some advisory companies recognize that their customers have an expectation that there is a digital platform they can use, so they wanted to have something that is fit for purpose. Some wanted to just have a tool that they can use together with their clients, so they can be looking at the same data together.

I had a company in Canada call me recently. They said the average age of advisors in their company is sixty-two years old, and they have no younger advisors coming up to back-fill. So, they said that they know they are going to have to find some other way to industrialize what they are doing. The guy said to me, “Frankly, we know that all of our advisors are not that great at giving advice anyway, so we need to figure out another way to do this.”

There are a lot of different ways this is taking shape.

A: We tell people the truth about their finances, such as, “You have no business being in the markets. You don’t have enough of a nest-egg, and you have expensive debt you need to pay down. You can optimize your 401k and get a lot more mileage out of that.” There’s a lot of things people can do before they ever get to the markets, which I think is one of the differentiating things about the Pefin platform. People have so many levers in their lives they can play with. They just do not understand what they are. Helping them understand is the core of the platform. For people for whom it’s appropriate, we will say, “Here is the asset allocation that we would recommend for you.” Then, they can invest through us, or not.

The other thing that I love about the way Pefin looks at investing is how it is integrated with your life.  Let’s say you want to buy a house and Pefin knows you are going to need $15,000 in a year for a down payment, but you have $100,000 that you are going to be putting in a taxable account. You have answered a risk-questionnaire, and you are risk level “8” on a scale from 1-10, meaning you want to take a fair amount of risk, you understand what that means and you have the ability to withstand some volatility. The platform knows that you may be a risk “8”, but you need $15,000 in cash in a year. Therefore, it is going to start taking down the risk on that portion of your investment to meet that liquidity objective. Whereas, if you are in a traditional robo-advisor platform, they are not going to know any of that about you, so they are not going to do that.  Pefin is very comprehensive and holistic in the way it considers what you said you wanted to do, and it is enabling that to happen.

Q: So, the best case is that I answer the questions and go through the steps; you learn about my life and tweak the investments on a minimum basis; and then, I just give you guys’ management over my money?

A: Exactly. It is very transparent and will tell you exactly what is going on. Right now, all of our investments are beta ETF portfolios. However, we did just have a new CIO come onboard, so now we are going to have alpha portfolios and alternative investments very soon. I think the majority of the population is going to be adequately served by the beta ETF portfolios. Nevertheless, there are people who are going to want and need more, so we want to make sure that we can service them as well.

Q: I want to say this is simplification. With that said, this seems like a smart version of the time, about ten years ago, when the Vanguard Mutual Funds would automatically cycle from 100% equity to 100% debt over a fifty-year horizon, which is a very crude metric. However, you guys are doing it with much more intelligence.

A: I think the important part is that investing is only one aspect of what Pefin is looking at in the context of a very comprehensive analysis.  The entire U.S. tax-code is built into the platform. Every state tax-code is built into the platform. Every university in the United States is built into the platform. Things like day care costs down to the zip-code are in the system. Thus, the amount of data that is in here is phenomenal. If any of those things change, the platform is going to update immediately and will send you a message saying, “Oh, by the way, this is what is happening for you. You might want to check this out,” or, “Here’s what we would recommend,” or something along those line. So, it is very focused on the holistic view of financial life. One aspect of it is investments, but that is just a piece.

Q: Let me ask you what you have learned specifically about the innovation process (i.e., running the company, being in the start-up). Concrete examples would be great. You gave an example of a big company versus small company perspective. I would like to know where the frontier of AI is, how to really operationalize that and make it useful, and the frictions that you face.

A: Sure. So, gosh there are so many.  Going into the point of the deterministic chat, I don’t love it, to be honest, but we know it is version 10. We know that is not where we want to be. In a perfect world you say, “Alexa, how are my finances?” And it would say, “It is great. Have a good weekend.” Right? But today everybody knows how annoying it is when you are talking to Alexa or Siri and they are like, “I do not understand that.” No one wants that to happen when they are asking about their finances, so we are still in early days of NLP.

In fact, somebody told me there’s a platform, which just came out with an NLP-based chat. The guy who was telling me about it yesterday said, “Every answer is: ‘we are still learning. We do not know’.” So, why would you do this then? I feel like there is so much tension between knowing where you need to get to and where you are – which is both the tension of a startup but also of where the actual industry and development is.  But things change really, really fast. Keeping that in mind, there are  still really good things happening in the industry.

You need people on the research side keeping up with NLP platforms as they get better and better. They’re mostly open-source, so we’re going to be building on those platforms. How do we continue to improve? Part of it is a resource allocation issue, part of it is staying on top of trends, and then a lot of it is finding out how to really listen to your clients’ needs. The art to succeeding is really in understanding what the analytics tell us.  We will obviously get an enormous amount of insight about what needs to change through the analytics, but in addition we actually spend a huge amount of time on the phone with our clients getting feedback. Sometimes there’s no substitute for just having a conversation with people and hearing what they want and need.

We have incredibly engaged users, which provide us with things that you would never get from analytics. For example, I was on the phone with a user whose family member had a stroke. She was saying that it would be really helpful if the platform could more easily explicitly capture various disability benefits, such as from social-security. That kind of information you’re never going to get from analytics. Without understanding people’s issues, they may just drop off from using the platform, go away, and you’re not going to know why. Thus, even as data-driven as we are as a company, there are instances where I still feel like trying to hear what’s going on with people and what they want is still an important dimension of making the user experience as good as it can be.

Q:  People want the small data sometimes.

A: Exactly, and you can’t underestimate it. Debt management is a huge issue for people, so obviously it’s a recent topic of discussion. We know that! Therefore, the platform has a whole guide on how to use, and we thought it seemed straightforward.  Nonetheless, we get feedback from some user like, “My issue is my student-loans. There’s no life event called ‘pay down your student loan’.” However, the platform explicitly helps people to pay down their debt. From our perspective, we thought it was obvious that ‘pay down your debt’ encompassed ‘pay down your student loan debt’. But, the user will say, “No, my problem is my student loan. I want to see, ‘Fix my student loan.’” That’s just a UI thing, so we can easily fix it, but it just underscores the need to meet people where they are.

A lot of the issue is the fact that we’re bombarded with information. The small, the big, the analytics, the future, all of these things. How do we allocate resources appropriately in the smartest way possible, and how do we know when we’re not getting that right? How do we make the changes that we need to make fast enough? I mean, as mentioned, the two-week sprint model is important for us to be able to quickly incorporate the urgent feedback we’re getting. At the same time, not everything you are going to try is going to work. So, you try it and it’s like “Well, that’s not great so try something else.”

Q: So, New York City is your natural home, right? Because NYC is the right place to be developing the fusion of tech and finance. So, from the talent question, I get that you feel like the tech community in New York is sufficiently strong now that it boomed. Can you disclose high-level numbers just so we can anchor?  You said you had 24 employees?

A: Yes, we have 24 employees now.

Q: There’s a stage in a company’s life-cycle when they can say it, I don’t know when that is, but you’re doing series A, right?

A: Yeah, we were doing series A so everything has been privately funded so far. It’s been a really interesting process, so I am definitely learning a lot!    

Q: Is there anything unusual about the team? Do you have a technical co-founder?

A: So, the founder, she’s a tech person. We were talking about it in the context of South by Southwest. I said, “I bet you we’re the only company being nominated whose current CEO and founder are both women. We also have a lot of female engineers. When you step back and you look at it, it’s probably is kind of unusual. I think seeing how many female engineers we have in the platform has given me reason to wonder, “Why?” If you look at sales and trading in banking, and other tech companies as well, they’re something like 90% male. So, why is it that a company you would expect to be 90% male isn’t? I think that having a female founder was very attractive to a team of strong female engineers. Then, once it gets going, it snowballs. It’s not like we go out particularly looking for women. We want to find the best people for the job. But, I think they seek us, like a kind of self-selection.

Q: And, I’m sure you get asked this question a lot, right?

A: Oh yeah, all the time. Having grown up in sales and trading, which is at least as male dominated as technology, it feels like I’ve been answering this question for the last 25 years. But, I think at the heart of it, you have to love what you do. I have been fortunate to work with phenomenal people throughout my entire career who really like what they do. If you like what you do, then it all kind of works out.

Q: Are you competing with Robo Advisers?

A: You know, it’s really interesting because the answer is “no” in the sense that, as I mentioned, that’s a feature that we have – but our platform is so much bigger. The real competition is human advisors, and the real question is, “How do you help people who’ve never had financial planning and advice understand they need financial planning and advice?” Not to overstate the case, but it’s kind of like when I first heard, “Oh, you can have a phone that has a camera and a flashlight.” I was like, “Why do I need a phone with a camera and a flashlight? I have a flashlight.” But then, you get it, and it’s like, “That’s so cool!”

Q: But, human advisors are going after that same population, right?

A: It’s funny, because from a scalability perspective, they really can’t. It’s just not cost-effective. I mean, as a human advisor, there’s only so many people you can talk to. That’s why they typically charge up to $5,000 to do a plan and also why you typically must have at least $100,000 in assets under management. Otherwise, it’s not worth their time. That’s why they’re super interested in partnering with us; we can bring them much further down into the funnel of a broader population that they can’t afford to service. I think it’s such a global opportunity. It’s really interesting, because in the United States we have a fairly strong culture of financial advice. It’s much more prevalent here than in a lot of other countries, so it’s a different issue or challenge in other countries where they’re thinking, “It’s not that I don’t have enough money to have a financial advisor, it’s just that financial advisors aren’t really a thing.” Nonetheless, financial literacy is well recognized as a global problem. I think that’s why people are kind of like, “I don’t understand this, and it’s stressing me out, so it would be great if there’s something that could help me deal with this.”

Q: That’s right. Is this your first kind of start-up?

A: Absolutely, so the whole thing is completely different. However, I will say that I’ve noticed a lot of Think Tech companies are very heavy on the technology and the finance in terms of appreciation, particularly if it’s a regulated entity. Having a real understanding of how that all works is actually pretty important in delivering the right product, really understanding how a regulator is going to look at all of this, and also knowing what a client needs. Technology is important, but it is just a tool. It’s really about finance and people.

Q: I’ve never hired a financial advisor. The last one I considered took me to lunch last year, and I kept thinking, “This guy is going to be out of a job within my lifetime;” and you know what, one of my best students got an A in accounting, and he worked for this guy in the summer. I just had to tell him, “Ryan, I just don’t think you should be working at this company. I just don’t think the human advising business is going to last.”

A: Exactly. However, I have a good friend who is a financial advisor at UBS working with an ultra-high net worth population. I don’t think her job is going anywhere because it’s all family offices and people with really complex trust structures and things like that. But, I was having coffee with her and she said, “I don’t think of myself as a financial advisor; I think of myself as a financial psychiatrist. I cannot count the number of conversations I’ve had along the lines of a rich old guy coming in and saying, ‘The kids are gone, time to ditch the wife and marry the mistress. What’s that going to cost me?’ and she spends time walking him through the many and various issues.” And that kind of consultation isn’t going anywhere, but it also applies to only a very narrow slice of the population.

Q: Now, what about macro trends towards passive investment? How does that affect you?

A: Well actually, I think using the beta ETF portfolios actually speaks to that. I also think it is the most appropriate way to go for most people. Therefore, I think that is going to cover about 90% of the needs people have. I think having alternatives, alpha portfolios, and other investment vehicles speaks to the desire or the ability some people have beyond a more passive approach. However, for the most part, it really should be beta portfolios for most people.

Q: So, you are not one to encourage stock picking?  I mean, the empirical and economic evidence is that stock picking is not really good for most people’s health.

A: I can support that not only empirically but anecdotally as well. I get asked all the time, particularly by young people, what stocks they should buy? First of all, my entire career has been commodities before I changed into emerging markets, so why would I know that? And also, as a young person you most likely have no money to lose. Unfortunately, I do think that there’s this mentality kind of like the lottery, so there is no strategy. I think helping people to understand that is not a good idea is important.

Q: You mentioned earlier something about Blockchain. Are you still working on that at all?

A: So actually, not with that same company, although I do keep up with it. We actually have a separate Pefin-sponsored project on the side we’re kind of working on. It’s really a non-profit we’re setting up. So right now, given everything else that is going on, it’s not as high of a priority. However, this project came from the fact that the entire KYC process and everything we’re doing to identify who we’re doing business with is really archaic.  Blockchain will certainly be important – and ubiquitous – in most businesses within my lifetime.