Transcript:
Penny Crosman (00:03):
Welcome to the American Banker Podcast. I'm Penny Crosman. Ethan Bloch is pretty well known as the founder of Digit, a fintech app that pioneered the idea of automated savings, which has become popular at several banks and fintechs. Digit was acquired by Oportun in 2021 and Bloch is onto his next startup and AI meets personal finance app called Hiro that launched in November. He's here with us today to tell us about his new company and app. Welcome, Ethan.
Ethan Bloch (00:31):
Hey Penny. Thanks so much for having me.
Penny Crosman (00:33):
Thanks for coming. So I just read on your LinkedIn page that you got hooked on finance when you were 13 and you day traded your bar mitzvah money. Can you tell us about that? Because most 13-year-olds don't even know what day trading is. How did you get into that and how did that work out?
Ethan Bloch (00:51):
For whatever reason, my first love was the computer. I had a computer, I was fortunate to have one at the age of six or seven and got to play with some of the original software on DOS and even Quicken, if you can remember that product. And then my second love was actually finance and I started buying stock maybe around the age of 11 or 12. My family had a stockbroker and I was like, what is this? Oh, let's try this out. My family has encouraged me to do that. And then I had a bar mitzvah in 1998. We can all remember 1998 was sort of the internet boom and the telecom boom, not unlike in some ways today. And very exciting. And I convinced my parents to let me take a large portion of my money I'd gotten from my bar mitzvah open up, I think it was an Ameritrade account or an E-Trade account, I can't remember which broker it was. And I started sort of day trading that volatility.
Penny Crosman (01:50):
Wow. I didn't even know you could open an Ameritrade account at 13.
Ethan Bloch (01:54):
Well, I'm pretty sure it was in one of my parents' names and I had the login and everything. And over the preceding couple of years I saw my money triple. That was very exciting. And then in 2001 in the telecom bust, everyone talks about the internet bubble, but the telecom bubble was much larger. I lost everything. And so by the age of whatever it was, 15 and a half, I sort of had my net worth wiped out. I say that laughing somewhat. It was a pretty good lesson to get early in life. And it sort of sent me down this path of saying, okay, I really didn't know what I was doing. I didn't have a sense of what a stock even is. And I had this curiosity and this question of, okay, what is an actual stock and what's a company? And that sort of led me down this lifelong even to now obsession with finance.
Penny Crosman (02:52):
Interesting. You've created several startups, Digit was one, but there were a few others as well, weren't there?
Ethan Bloch (03:03):
For whatever, again, reason I've always started what I called back in the day projects, and that's just what I did. And it wasn't later until I went to the University of Florida where I learned the word entrepreneurship and I initially didn't like the word. I was like, this is stupid, it's just doing projects. And I'd started a number of businesses and when I moved to the Bay Area around 2008, I started my first venture backed business. It was a product called Flowtown that was software for small businesses to market themselves on Twitter and Facebook. And then after that company was acquired is when I started my journey towards Digit.
Penny Crosman (03:42):
So what inspired you to create Hiro?
Ethan Bloch (03:46):
Hiro has been a dream I've had for around 15 or so years. And so just rewinding a little bit, when I sold my first venture-backed startup Flow Town, I had sort of made enough money where I could do whatever I want but not too much where I didn't have to do anything. And so I went in this place of trying to discover where am I going to spend a large portion of the rest of my life? And I slowly came back to finance and specifically personal finance and had this big question of how can I help people with their personal finances. And around the same time, I had discovered this short story written by Orson Scott Card called the Investment Counselor. And it's a short story that takes place in the Ender's game universe and in the Investment Counselor there's this AI named Jane.
So for those that don't know, Ender — spoiler alert — saves the universe in that first book and then starts traveling around the universe at light speed and accrues a really large tax bill and doesn't know how to pay it. So he tries to get off and stops by customs and they're like, you owe this huge tax bill. He is like, I have no idea to file my taxes. So someone saved the universe and doesn't know how to file their taxes. It's such a great metaphor to modern life. And Jane, this AI that shares her backstory, she sort of presents herself to Ender and is like, I can take care of this for you and solves the problem for him. And I sort of had this desire, everyone should have a Jane, everyone would be so much better off if we had this brilliant AI who worked just for us, helped us sort through all the financial stuff we need to sort through in life and took care of it for us.
(05:33):
It sounds amazing. And that was actually the spark that led to Digit. And in Digit's case, we were trying to build this AI to rewind the clock. And this is around 2014 when Digit launched. It was actually a chatbot. We were the first financial chatbot in the market way back then and customers would talk to about different financial needs. And what we ended up discovering pretty quickly was helping people save was a big opportunity. And so that's really where digit focused, how can we help people build more savings? And we ended up stumbling on algorithmic saving, studying someone's checking account, moving small amounts out every day and actually adds up to thousands of dollars a year that we could set aside for people that they otherwise might have spent. And at the same time, we started to really hit the limits on what was possible with the technology that day in terms of conversations and neurolinguistic programming and what do people want to talk to digit about and what can digit actually do.
(06:34):
And the most common response a lot of people would get is, I'm sorry, AI can't do that. So while we could handle basic inquiries, what's my balance and how much have I saved? And when someone's like, how do I think about opening this retirement account in a certain way we just couldn't handle it. And we ended up sunsetting the whole conversational aspect at Digit later on, maybe around 2017 or 18 or so. And I thought, okay, we really tried to make this conversational helpful ai, but that interface of chat just is it there? And maybe there's other ways to do this through just automation. And that's the eventual path. We kept follow with Digit, we automated savings, then we sort of automated paying down debt and then we helped automate actually setting aside money for bills and so on and so forth.
Penny Crosman (07:23):
So you sunsetted the chatbot in Digit, but you kept going with Digit as a savings app.
Ethan Bloch (07:29):
Exactly.
Penny Crosman (07:30):
And then ...
Ethan Bloch (07:32):
Yep. We ended up helping a little over three and a half million people save $9 billion with Digit. We eventually were acquired by Oportun and we, even today we became sort of Oportun app and their savings product. It's still in the app store, it's still a highly rated savings experience that we're really proud of and found a great home with Opportune in helping their business. And I left Opportune in 2023 and I thought, okay, I'm probably done doing startups. Digit was a 10 year journey and during the same time I built my family and I built digit into the sort of success and outcome that it was and it was time for me to take care of myself and take care of my family. And so I left around February of 2023 I think.
(08:23):
And on the same day my main partners Doshi left who at that point was the interim CTO at Oportun. And I went on vacation more or less and Roush dove into large language models because just a few months prior chat GPT had come out and when it came out in 2022, it was not obvious to me yet at all that that would be a new unlock for some of the stuff we were trying to do at Digit. The original chat, GPT was good, but it wasn't exceptional and he certainly wouldn't trust it with your money. And so Rouge went on his exploration and learning and educating himself and I was taking a break and spent time with my family and managing my money and we would keep in touch and Roche's ideating different potential ideas. Fast forward nine months, we're catching up and he's showing me some of the things he learned and is doing. And we both had this, I can't even describe it, sort of like the universe calling moment, which is like Holy cow, you have to go see now if the dream we had at Digit is possible because of large language models.
(09:38):
And that was when we had started Hero and this is now 2024 or at this point, early 2024. And that was it. That was sort of the mission. It was, okay, hero is going to be this exploration way out on the frontier at this intersection of a large language model and personal financial data. And the question was, are there things that we can now do for people that are super valuable that just were not possible before large language models.
Penny Crosman (10:12):
That makes sense. So how did you go about it? Did you start testing some of the different large language models and feeding them financial information and seeing how they did?
Ethan Bloch (10:24):
Yeah, so initially we built a small team of people we have worked with before that we can move really fast with and who are exceptional at their discipline and craft. And we started trying lots of different things like you just said. So one of the first things we did is, okay, we need to ingest financial data and so let's go do that. And thankfully it's easier, although still somewhat contentious, which is another topic we can talk about with accessing your data at your bank and pulling that into Hero. And you could ask the LM about that data, okay, how much did I spend on this or how much did I spend on that? And what, that was one of the early prototypes. But from there we started to think about also actions. And so we prototyped something that could log in for you on your behalf in your Chase account and pay a bill for you. And that would be LLM driven with some robotic process automation, I dunno what the exact number, but we probably had 20 different things of products or value props we were trying to deliver that were unique because of the LLM. And through that exploration we ended up realizing the most value we could deliver right now is actually in the wealth management discipline.
(11:47):
And if you think about wealth management, there are these sort of what we call these high impact workflows like building a financial plan, allocating your portfolio, doing tax planning for this year or next year. And it turns out these things are now achievable with individuals using an LLM and a purpose built piece of software where previously they would've had to go to an investment professional or some professional service business or they're just Googling and doing spreadsheets and hoping that they're doing it right.
Penny Crosman (12:24):
Did you start building your own model for this or did you take an existing model and start tweaking it?
Ethan Bloch (12:31):
We use, I hesitate to say off the shelf, but we use other foundational models for different based on what's the right use case. And so we'll use the latest GPT 5.1 previously GPT-4 0.1 from OpenAI for a bunch of things. We'll use actually Gemini for some things. And we're constantly testing. We've built Hero in a way we can quickly try a new foundational model and run it against our evaluation suite and see how it performs to know if we should replace a different model we're using for a certain workflow or task.
Penny Crosman (13:10):
And do people have to link accounts or allow data sharing from their bank accounts for this to work?
Ethan Bloch (13:18)
They don't have to, this is actually something we unlocked, which provides someone getting used to hero or trying it out. A lot of value for that to connect anything. And so what you can do today, today Hero is very focused on financial planning, has several other features, but that is the core experience that's in the market today. Anyone can sign up, it's totally free right now it's in this public beta phase. And when you sign up, hero will sort of walk you through setting up the financial plan and asking you questions. You can describe it, your income and your expenses, and you can describe it, your assets. You don't need to connect anything and it will plug those in and model them correctly. And then very quickly after just 30 seconds, a minute or two, you can sort of forecast your net worth over the next 45 years.
(14:08):
And see what financial trajectory you're headed on. And then very quickly you can say, okay, well what happens if I buy this house or that house and see the difference in the outcome and dig into the actual what it does to your income and expenses or you're going to move locations or you're going to change jobs or any of these big financial decisions we're all left with on our own or maybe with a spreadsheet we can now get the help of a really purpose-built AI to think through that situation and that circumstance.
Penny Crosman (14:40):
Now you mentioned earlier the challenge of getting people to trust a chatbot with their money, financial information. How do you get people to divulge all those data points you were mentioning without fear of some kind of privacy or security issue?
Ethan Bloch (15:02):
Yeah, so we think about this on a few different dimensions when it comes to privacy. We built hero from the ground up and we're not a bank, but in sort of a discipline of a Swiss bank where privacy is the utmost key, right? You're going to come to Hero, you're going to share intimate details about your financial life, what are we doing from the ground up to make sure that information is secure, to make sure it's encrypted, to make sure people on the team can't just see it if they want to. We actually have a max privacy setting and hero that if you turn it on, you just completely disappear from our internal systems and we can't see anything unless you write to support and explicitly share information with us. We can, because we're not the foundational model ourselves, we can make certain guarantees to the user that their data is not going to be used for training and won't be leaked to a model that can use it for training.
(15:56):
And we still have more work even there where we'll get to a place probably running on our own infrastructure, hosting our own models completely to be able to fully back that guarantee and stand up to that as hero continues to grow. And so those are some of the things we think are essential again, for someone who's going to be an AI wealth manager to be able to show and guarantee to the user. But then there's other things which is, okay, I'm doing financial work with this. This is something we learned early on. How do I know it's right.
(16:27):
How do I know this isn't just vibes? And because LLMs are probabilistic, you'll get a lot of times a different answer. So how much I spent on DoorDash last month? You'll get a slightly different answer potentially if you give it data. And it's doing that math all at inference time. And so what we've done, depending on the different task at hand is the LLM is typically never doing math inside Hero. If you're asking in how much you spent last month, it has a tool, it's querying a database, the code is giving it the number determined that they stickly and then Hero, the LLM is giving you it back. And that's a simple example, but when it comes to financial planning, the whole plan itself is a deterministic engine that's been built for an LLM to operate. But when you're forecasting your net worth that's happening in deterministic code and it says your net worth will be $10 million in this year, refresh the page a hundred times, it doesn't change. And that number is given back to the LLM and the LM just shares that number with you. And so we call that verifiable math and it's going to be essential for anyone that needs to trust in LLM to do anything high stakes involving math or finance. You're going to need to make sure the numbers are actually verifiable.
Penny Crosman (17:40):
Sure, that makes a lot of sense. And when you mentioned that short story that you had read an Understanding Investment counselor. Yeah, an understanding of taxes and the tax code was what that bot had to offer. Have you built in information about the US tax code into this?
Ethan Bloch (18:00):
We, yeah, we built a mini TurboTax inside of the forecasting engine that is US tuned today. And so when you're in there and you're detailing different tax specifics, single married filing separately jointly your zip code, it's using a tax engine to figure out what not just your income taxes will be, but what your investment taxes will be and what your contribution limits are and all those sorts of things.
Penny Crosman (18:35):
So you mentioned this is a public beta. For how long will it be in that public beta and then do you know what you're going to charge when you start offering it?
Ethan Bloch (18:46):
Yeah, so we don't know yet exactly what we'll charge or how the business model will work specific to subscription. There will be a subscription. We do know that much. We think that's one of the right ways for a product like this to make money. So we have financial planning that is out. It's sort of a cornerstone to hero. It is a cornerstone to people's financial life. And we have a couple of more of these sort of what we call high impact workflows that we're actively working on. I think when we ship the next one, maybe when we ship two more of them, but maybe just the next one, we'll also figure out what that first pricing will look like and that will go along with it. And I really hope we get to a place where Hero comes out of public beta. It's a very clear subscription sometime in the next six months.
Penny Crosman (19:42):
Okay. Now I'm seeing other announcements of partnerships being formed between personal finance providers, fintechs and foundation model providers. For instance, Intuit announce their partnership with OpenAI to let people use chat BT to kind query their finances. How do you see an app like Hero as sort of different or better?
Ethan Bloch (20:07):
Yeah, so I think it's hard to fully understand what the Intuit OpenAI partnership will look like. It's just words at this point. And Intuit needs to figure out their AI strategy. And so this is one of the things they're doing, which makes a lot of sense how that interplays what customers actually do, how useful that is. What takes place in open AI's world versus Intuit's world. This is a big question mark right now. I think there's definitely a lot of opportunity to, I want to connect TurboTax OpenAI and asking questions about my past tax returns. Obviously I should be able to do that Intuit, but hey, I'll do it in OpenAI as well, if that's the way it's set up. I think that can make a lot of sense. And same thing you can imagine integrations with QuickBooks can make a lot of sense in terms of what we're doing at Hero right now in the US there are millions and millions of higher income households that don't get great financial advice and they don't get it because there's not enough wealth managers in the industry. There's actually a significant shortage of people. And so it's really hard to get great advice on your investments, on your financial plan, on your taxes today. And that is the opportunity that Hero is suited for where very soon these millions of people, tens of millions of people who don't have this trusted Wealth manager advisor are going to be able to have one. And we're very focused on that specifically that for the foreseeable future.
Penny Crosman (21:45):
Yeah, I was going to ask you who your target audience is, and you just told me how will people find out about Hero, do you think? Are you going to do a lot of marketing?
Ethan Bloch (21:54):
I think the way it is, we learned a lot about how to market a consumer product at Digit, and we will borrow some of that into Hero. Now, what's different with Hero is it's a much bigger, more important relationship. And so people trusting Hiro and telling their friends about Hero or their family about Hero is going to be even more important. And so that is the way for the foreseeable future will continue to grow, is just continue to deliver an experience that's just blows people away. And so they tell other people how great this is and that through that can proxy trust about Hero.
Penny Crosman (22:40):
Sounds like a low cost customer acquisition approach. And if someone is listening and they work in a traditional bank, do you see a possibility of partnering with traditional banks?
Ethan Bloch (22:55):
It's hard to say. I think definitely possible. It just the devil's in the details. How is it work and what's distribution look like and what do the economics look like and such. But if there's someone in banking, a registered investment advisor at a wirehouse or somewhere and they want to figure out how do you scale fantastic client experience for the mass affluent and some of the lower high net worth customers, we have figured that out.
Penny Crosman (23:28):
So you mentioned that this trial period will probably be about six months. Is there anything else on the drawing board for Hero in the coming year?
Ethan Bloch (23:36):
It's really just for 2026 for us is filling out that suite of things that when a client looks at Hiro, it can do most of what they need from a wealth manager. And so whether that's their financial plan, their portfolio, and how that's allocated across their life, their tax planning, some of their estate planning, that's the goal for us. We want to try and fill that out. And not all of this is solved yet. There's still things we need to invent to do that because we're right here on the frontier figuring out how an LM makes this possible. But that's what we're focused on certainly into 2026 and beyond. And then starting to build our brand so we can be a trusted brand and name in the space. And that's going to be our focus through about 2026.
Penny Crosman (24:26):
Alright, Ethan Bloch, thanks for joining us today and telling us about the business you're building. And to all of you, thank you for listening to the American Banker Podcast. I produced this episode with audio production by Adnan Khan. Special thanks this week to Ethan Bloch at Hero Finance. Rate us, review us and subscribe to our content at www.americanbanker.com/subscribe. For American Banker, I'm Penny Crosman and thanks for listening.