Stephen: If you are running a payments company, a crypto company, a traditional bank, this is probably the podcast you wanna listen to because we talk with Joel Robinson, who's the CEO and co-founder of Hummingbird. They are helping companies create a dashboard for all their CRM purposes. Think the sales force, but for your customers.

To make sure that they're not committ any financial crime or have any irregular behavior on their account, including things like account takeover and fraud. Joe goes deep into some of the situations happening, especially with stable coins and cryptocurrency. We talk a little bit about the emerging trends in financial crime.

He even gives us the low down in discussions around what it's like to be a CEO and what keeps 'em up at night? This is a really fun podcast, and he even gives a couple huge book recommendations for any founders in the payment tech or crypto space at the end of the podcast. Hope you enjoy it.

Stephen: This is your host, Stephen Sargeant, the Around The Coin podcast. We have a favorite episode 'cause you guys know I love talking about crypto and compliance, and we have the co-founder and CEO of Hummingbird, Joe Robinson. Joe, tell us a little bit, I wanna start off with a little bit about what Hummingbird is right off the bat.

Then I'm gonna go back into your background because I'm super interested. I know a lot of people have heard about Hummingbird, so your company's done a lot of great work about bringing awareness to the company. But I think if I asked somebody exactly what you do, they would struggle and I always want people to know exactly what the companies that we bring on do.

So why don't you tell us exactly what Hummingbird is and how it came to be.

Joe: Yeah. Well thanks for having me on. I'm happy to chat about it. Hummingbird is a platform for investigating financial crime and fulfilling regulatory obligations that, uh, banks and credit unions and payment companies and crypto institutions and things like that have. Um, I like to draw an analogy for folks that are not in our industry to like a Salesforce or something like that.

Um, except instead of sales prospects or customers, the, the folks that are in that CRM are, uh, your own, um, customers of the financial institution and your, your fulfilling obligations that you have around investigating financial crime. So we're really proud to be in that space. Our mission is fight financial crime, and, uh, that's all we do all day.

So,

Stephen: I love that. And now I think about like Salesforce. They're recording every message, every conversation, every time you reach out to somebody on social media. I can see now how that makes sense for the customers of your customers that are using their services, but they're also have open source information.

You're pulling all that data and all that risk and I'm assuming, and creating like a risk scoring. Would that be a simplified version of what you do?

Joe: yeah, you can, yeah. Let's, uh, let's actually use the analogy if, if I was, uh, investigating you, Stephen, um, we'd probably be looking at your profile information, some of which you submitted when you signed up for your credit card or whatever. Um, we'd be looking at activity, we'd be looking at, um, your interactions with the app, your transactions, things like that.

Um, might do to your point, you might do some open source intelligence as well, so, um, you know, Google your name and try to find associations and things like that that you might have online. Um, and at the end of the day, um, you know, we're basically trying to understand if, uh, behavior, uh, you know, your transaction behavior and things like that is normal or if it might be indicative of organized crime.

Uh, fentanyl rings, um, you know, lots of different stuff. Corruption, bribery, um, any sort of illegal activity that leverages the financial system.

Stephen: You know what's interesting about your resume and your background? You worked at both Square and Circle on very early days. Um, especially when you think about like your product manager at Square, I think it was between 2011 and 2014. Were you already interested in crypto at that stage or blockchain?

That's fairly early even for crypto status. Uh, what were you doing at Square? Was it strictly on the product side? And I'm assuming if you're working in product management, you're talking to developers, you know, maybe blockchain or at least Bitcoin comes up maybe in one or two conversations.

Joe: I, I was, uh, not even remotely aware of our practice area at that time. Um, you know, even before the financial industry, I had a whole career as a product manager in the online video space. Um, and, uh, I joined Square, yeah, it was a very small company. In 2011, I would joined as, uh, one of the early product managers there.

Uh, the first product I worked on was, uh, the merchant dashboard. So the, you know, the kind of behind the scenes view of the merchant account and, um, how they were doing sales wise and accounting and things like that. And, um, a little bit after that project, I actually started the Square Online Stores product, which is their first e-commerce product.

Um, actually, uh, I worked on that as the product lead and my co-founder and our CTO Jesse Reese was the engineering lead. That's how we met. Um, that was probably my first, um, my first dive into the pool of financial crime. Uh, as we were spinning up e-commerce, you know, we realized that E-commerce marketplace actually a really great place to launder money.

If you can be the buyer and the seller, um, you can just, you know, sell yourself fictitious goods or things like that to launder money. And, um, I got a real education in what the practice area looked like there. Um, I know your podcast about crypto fun. Fact is, uh, the Square Online Stores product in 2013 is actually the first place that Square, uh, worked with crypto.

That was my team. Um, and we had a, a, a, I think a hack week project to accept Bitcoin as a payment type, uh, for that e-commerce store. Um, so I learned about that as well.

Stephen: That's super interesting. But then you dive into circle and I know anyone listening to this podcast is like, oh well duh, that's an obvious move. But this was like almost a decade ago, you know, stable coins were not exactly where they are today with the hype, especially after the US Trump administration has come in and really basically baptized the use of stable coins globally.

What made you decide to jump in, you know, headfirst into crypto? 'cause back at that time it still seemed very risky. People were not talking about crypto. It was a little before I jumped in and I remember when I jumped in Bitcoin was at like six or $700. So you were really early in the stable Coin era.

Joe: Well, and, and to clarify as well, I, to my knowledge, so this is 20 13 20 14. I, I was talking to 'em in 2013 and joined in 2014. Um, they, I don't think stable coins existed as a, uh, um, a concept yet. Um, so this was still really where Bitcoin was the, the main show in crypto. And, um, you know, to put it in context, folks know about like the rise of financial crime in that area through, uh, silk Road and Ross Ulbricht and things like that.

That was all 2016, so that was actually two years later. Um, yeah, I, you know, uh, the founder of Circle, um, Jeremy Aire, uh, it, I think that was his third or fourth business. His, uh, prior venture to Circle is a company called Brightcove that also iPod. Um, I spent four years at Brightcove, working with Jeremy there, and, um, I, I would have to give him credit for pulling me into the crypto space, uh, to join Circle as the VP of product.

So.

Stephen: So you worked at two companies that had successful IPOs, um, and then you co-founded Hummingbird a short time after leaving almost like nine years ago. That's a long time ago. Like how has Hummingbird evolved from back then until now? As you said, even said crypto wasn't, you know, the main part. Stable coins wasn't the main part of the conversation.

What were you working on in the early days and how did you transition to now where you're servicing crypto exchanges, fintechs, uh, you name it, uh, hummingbird is working with them.

Joe: In, um, yeah, and, and you know, some parts of Hummingbird over the, uh, I think we've been in business for eight years. Um, some parts have not changed at all. Um, our mission is exactly the same. Our value system is exactly the same. Um, and to some extent, one of the products that we're known for was the first product, which is, uh, suspicious activity reporting, uh, uh, which is a, a subset of the anti-money laundering practice area.

Um, other things have changed, of course. Um, you know, just team and capabilities and the breadth of the things that we're able to do now. Um, you know, the, the, the story of the company is about two years of r and d on suspicious activity reporting. We were super blessed with early customers and former colleagues often that really believed in us.

Um, the very first company we signed was with rippling, uh, the crypto exchange. And then we, we caught Brex, um, I think when Brex was maybe around a hundred people. Um, but, um, you know, and then we had a third customer I can't name that was a large enterprise customer and needed help with their SAR filing.

Um, so that, that is where we started. Then we started to take on more of the investigation work. Then we started to broaden into all the myriad workflows and practice areas that risk and compliance teams cover. Um, which are, you know, anti-money laundering, anti-bribery and corruption, customer due diligence, uh, you know, collaboration with law enforcement.

So subpoena response as it's called in the practice area. Um, customer due diligence, all sorts of stuff. So now we just try to be the, you know, the place where those teams do their work, uh, regardless of the subject matter.

Stephen: That's super interesting to start with sars and for people that don't understand su suspicious activity reports, let's just say a basic example. You go into a bank full of duffel bank, you know, full of money that might be smelling like marijuana. You go to deposit that money, even if the bank accepts it, they may do an investigation internally and decide what you're doing is, you know, potentially due to either criminal activity or terrorist financing.

And they may file a report to a financial intelligence unit or to a regulator stating that they believe that this customer or the person that's associated with this customer may be suspicious of committing some kind of money laundering. So you have to file these reports. What made you start with that?

Although I've worked at the bank and the crypto exchange, I know how cumbersome filing these reports are, how much administrative time it takes away from teams. It, it's hard to think of somebody that built a whole business around it. However, there's, you know, full departments that were dedicated to just that job, so now it makes sense.

What problem you were solving.

Joe: Oh yeah. And, uh, you know, our, our, I think Hummingbird was one of those entrepreneurial ideas where we were really solving for, for problems that we had as operators. Um, so part of our founding team, uh, myself and Matt Van Buskirk worked together at Circle. Um, over time I transitioned from the head of product at Circle to the head of risk and data science.

Matt was the director of compliance. And so you kind of see those two practice areas that we serve now. Risk and compliance start to emerge. And, um, you know, this is early days and we were looking around the market meeting with a lot of different vendors trying to put together our tooling and data sources.

Um, you know, had the benefit of seeing what was out there and what was not. Um, and there was just nothing out there for, um, internal kind of case management orchestration, filling out these filings, which is extraordinarily cumbersome. Um, you know, audit, audit trails, things like that. Like, like a lot of institutions in those days.

We were kind of trying to keep track of the compliance program and some of the work that we're doing through spreadsheets and file folders and things like that. You can totally do that, um, until you start, start to scale up and then you're gonna have a serious headache about, um, trying to maintain audit trails and, and records and things like that.

So, um, I left un unrelated to Hummingbird. I, I left Circle. I wanted to do some other things. I was living in Europe at the time. Um, I spent some time at IDEO doing some consulting for a large bank. Um, and then, uh, Matt, you know, and I reconnected like a year and a half later after our experience at Circle.

Um, uh, Matt is a interesting, he's a former regulator, so he's a former bank examiner for the matters that we talk about. He is got, you know, some of the deep, deepest subject not, uh, subject matter expertise in the world on these topics. Um, and I'm a mostly a product guy and, um, you know, the team came together and, and, um, we started to work on, um, you know, building a better way to do these things.

Stephen: If you had to, like, I, I could only imagine the amount of time and money that you're saving some of these companies, as you said, that are scaling at an enterprise level. If you had to put a number, is there a certain amount of time or amount that you believe that you're sa saving the average company by not having to employ 10 people simply to write SARS and submit them?

Joe: definitely. We, we have ROI studies for, you know, not just sars, but different parts of the, the practice areas that we solve. And it kind of depends on, you know, which financial institution we're talking to. 'cause like what they're doing might be fairly tailored to their own business. Um, but I know for, you know, uh, time series studies and things that we've done with SAR filing, like, um, we took one fairly large, uh, customer in the payment space from about 180 minute handle time per sar.

So about three hours. Um, to less than three minutes. So you're looking at just a massive efficiency gain through, uh, better automation, better quality control, um, a lot of data transformation and, and validation that we do. Um, and then wrapping the whole process in connectivity to the f IUs that you mentioned earlier, the, the receiving, uh, body of the government.

Stephen: I think what companies don't realize crypto changes in fintechs is that it's not just a time it takes to do the actual task. Because it's not really using that person, which is usually the investigator's core potential. They're doing redundant, you know, mediocre work. So they're gonna get distracted.

They're gonna want to go to, for a coffee halfway through. They're gonna wanna look at something more exciting halfway through. I think that's even hard to do with time studies. 'cause you're basically saying do the task until end. But I think because it's such a long and arduous task, you don't realize how much distraction gets implemented by some of the people because they're not really doing what they are their best to do, which is investigations and critical thinking.

So I don't think companies take that in mind too, when they're doing some of these thought processes between, you know, getting an automation tool and a significant tool like yours leveraging AI and doing it manually. I don't think there's many time studies based on how people, how easily it is for people to get distracted.

Joe: Yeah. Yeah. And, and I, I don't know if people, one of the things I've learned about Hummingbird over the years is that if you're outside this industry, you probably don't understand the scale of it. Um, you know, I think most of us have an image from Hollywood and movies and stuff of investigators that are spending several days or a week with a case or whatever, and they're out in the field and doing all this research and stuff like that. That of course is true for, um, field work that law enforcement does. And, um, you know, and, and some of the most involved cases, but, um, actually a lot of investigators sitting in banks and things like that are, are trying to get through five to 10 cases per day. Um, and, you know, the, the amount of time that they're able to spend on each is, is, um, resource constrained.

Uh, and there might be, you know, at a large bank, you're probably looking at three to 5,000 investigators that are doing this work as well. So, um, it gives you a sense of the magnitude of, um, how big this practice area is.

Stephen: I think by the looks of it and by your website, it looks like you scaled a lot of those processes that you're talking about using ai. Now AI is always very tricky when it comes to regulators. You know, there's a debate about it being a black box and you don't know about the decision making or the reasoning of the ai.

Can you talk about how you've been able to help clients scale, uh, using automation and ai, but also satisfy the regulators? Like you're talking Matt, who he used to work for, how you help satisfy the regulators to show that your AI is working. But you can also document the process on how it got to the final decision.

Joe: Really two, two modalities of AI that are pretty useful in this practice area. The first is, is writing, um, which of course everybody is familiar with for large language models. Um, there, there is a, a communication process. What a suspicious activity report is effectively a communication from the financial industry to law enforcement.

Uh, those reports are critical evidence for prosecution. So prosecution will use those written reports to substantiate, uh, things they're observing in the field. They might see a drug cartel that has multiple people, uh, selling on street corners, but then they're able to connect. Those activities and corroborate the evidence or back up the evidence, uh, through the written report of their financial transactions as well.

Um, and so there's a big part of like, literally just writing up the report. Um, AI is tremendously helpful there when, you know, the investigator or the, the bank sort of knows what they want the report to say, but then it was still time consuming to have hands on keyboard and come up with the exact PS to say it.

Um, obviously AI's really good at that. And then it's, it's a superior writer. Let's not kid ourselves. It's a better writer than I am. So like it's, it's gonna write a clearer, more understandable report that law enforcement can actually use, assuming it knows what you, you wanna have in the report. Right. Um, and that brings me to the second modality for AI in this practice area, which is they're, they're looking across lots of different information, activity, support tickets.

Um, you know, transactions, messages you might include with transactions and peer-to-peer apps, things like that. Um, you know, reports about you in the news, things like that. And so AI is, uh, particularly good at dropping in all that information and creating summaries and then looking for the parts and prompting for the parts that might be useful, uh, to an investigator.

So those are the two ways we use it. Um, they're very, very effective at, um, assisting, uh, this practice area. Um, and I think we'll see more and more of that in the, the coming year.

Stephen: I think for anyone that's not done crypto compliance or even traditional a ML investigations or compliance, you don't really realize how hard some of the SAR reporting is. And it seems to be a contentious point for regulators 'cause they're saying, Hey, people aren't running very good SARS for us to go out and get the bad guys the way we're supposed to by disseminating that information to law enforcement.

Uh, and I saw in practice this, I saw on the website filling out this arm like, oh, this is way better than a lot of the SARS I was reviewing, uh, when I was doing, you know, quality control work. Uh, and its consistency. I think, you know, working with regulators for a long time, especially in the crypto space.

One of the default judgements that, you know, regulators always had against many of these international, fully remote teams is that they weren't consistent in the way they wrote the sars or the way they wrote any documentation. And, you know, if one person's trained in one country and the other person's in another, like there was no consistency.

And by having a tool like this, I can see how this would make at least, you know, the formulation of these sars a little bit more consistent.

Joe: Yeah, I, you're right. I mean, uh, regulators will take that lack of consistency as sort of like an early warning sign that maybe your training program and, and your overall program isn't super strong. So. Um, this certainly helps ensure, you know, and the other thing AI can do, of course, is be a, an assistant and a and a coach, um, you know, you can feed ai, the desired instruction manual, if you will, for investigators.

And AI can help them understand what they should be looking at, what they should be paying attention to. Um, it could flag things that they may not otherwise see in the investigation data. Um, so net net, like the, the quality improvement, the speed improvement, the efficiency gain is, um, really powerful here.

Yeah.

Stephen: You know, for companies like stablecoin companies and payment processors and payment companies, we have a lot that come on this show and that listen to these episodes. They're now adopting stable coins and cryptocurrency, but they're also now starting to build out their own A ML programs and anti-money laundering and compliance is very foreign to them.

What are some of the biggest benefits that you think your platform will bring to these newcomers that are now starting to file sars or you know, report on a anti-money laundering issues?

Joe: Just starting with a topnotch program. Honestly, I mean, every financial institution in the world, um, that, that is a legal financial institution, of course, has to have a program like this. And, um, you know, you want to be sure that A, um, you're doing a great job so that b um, compliance doesn't get in the way of your business.

That, that's sort of the goal, right? Is, is to be sure that you're playing by the rule book of financial, uh, licensing, uh, that you're stay compliant and that, um, you never have regulatory issues that get in the way of your business.

Stephen: Do you think? You know, when I was working on the crypto exchange, the toughest part was transitioning from like fiat to crypto. That whole currency, bridging and moving in between stable coins now is harder to now see the, you know, secure or the least a source of funds. 'cause it's moving from fiat to crypto so quickly.

Is this an industry-wide problem? How do we solve this? You know, with stablecoin adoption going through the roof.

Joe: I think it is an industry wide problem. It's sort of, um, I, you know, I think, um, compliance people don't necessarily tend to view crypto or digital assets as all that different from other financial products and services. I mean, we already have like many different currencies in the world. Fiat currencies.

We have, uh, debit cards, credit cards, loans, uh, bankers checks, cashiers checks, um, you know, you name it, like, there's, there's a, a plethora of different pro financial products out there. Um, yeah, I think where digital assets get tricky is they can move around the world, of course, in real time. Um, they are getting popular, so the counterparties that you're interacting with might be questionable.

They're interoperable. So, you know, you can, you can send between different exchanges and things like that without those exchanges having a prior relationship. 'cause you can leverage the blockchain, which is open source. Um, and um, and I think, you know, there's no magic about stable coins for criminal activity.

I think criminals appreciate a stable currency as much as, uh, the rest of the world does. Um, but um, you know, it certainly adds a degree of difficulty, I think to, you know, doing the work and, and a degree of risk candidly, um, where, you know, you really might not know where that money's coming from unless you've done some, some diligence for it.

Stephen: You know, you must see a lot of trends, especially when it comes to financial crime across the crypto exchanges, fintechs, and. What are some of the biggest compliance challenges that we should be looking out for going into 2026 as we're already in 2026, but what are some of the trends that you're seeing that we should be keeping out?

And I know fraud is a huge conversation piece now. What are some of your thoughts, things that you're seeing either with your clients or, you know, feature, feature requests that your clients are adding you, asking you to add into the, the entire technology stack.

Joe: Well, yeah, I mean, I think, let, let's talk about crime first, right? Like, um, I, I've been in the, I've been in or around the anti financianal crime world for about 12 years now, and, um, I think even over the span of my career, I already mentioned some of the dark web markets were an early concern, um, around 20 15, 20 16, 20 17.

Um, they are bigger than ever today. Um, I think most people don't realize that like the dark web, uh, marketplaces are still actually quite active and very large now, um, the other thing that's a really pronounced change is one that, you know, your, your listeners will recognize from the news, like scams and fraud, um, are now, you know, being, uh, committed by almost like nation state size actors, right?

Like, um, we have these, uh, known scam centers that are almost the size of small cities in parts of the world, and, and the people working in them are in a form of modern slavery. Um, they're held there against their will, they're beat at night and things like that. Um, and, um, you know, that is where some of the scams, the elder expectation and things like that, um, pig butchering, uh, those are coming from those highly, highly organized.

Um, uh, you know, sort of, uh, uh, companies, if you will. So that's driving change. Um, you know, I think another thing, uh, that, you know, folks are already aware of, but we need to be even more aware of and vigilant of is AI is a great weapon for fraudsters, right? Like, if I can personalize my attacks and put them in your language and make the language sound like friends and family that I might be impersonating, maybe even use DeepFakes to send you videos or video recordings of your kids, things like that, um, you know, that's all happening, right?

And, um, I, I actually think, uh, sadly it's probably early days of that we'll see a lot more of that, um, over time.

Stephen: You see right now there's kind of that Wolf of Wall Street thing where a person's doing hand gestures and they have Leonardo DiCaprio from Wolf of Wall Street following exactly what they're doing in real time on AI that's already getting scary. It's, you know, non face-to-face onboarding is already tough enough, even with tools.

I think that's getting people a little bit more scared, but maybe even not to breach enterprise level cybersecurity and onboarding. But definitely the fool of grandma that, you know, hasn't seen their chi their grandchild in a while and is calling for money. Super easy to fake those with the tactics that you're talking about, which is like this big pig butchering scream, uh, big pig butchering, uh, scam that's going on around the world.

Billions of dollars are being siphoned, uh, by the victims. I'm curious from like a, just a strictly UX point of view, like the reporting to regulators. Looks like we're using fax machine type forms. Like do we need a, a redo, do we need a, do we need a revolution, a reform when it comes to UX and treating maybe compliance like a, almost like a customer consumer facing product.

Do we need some work there? Any thoughts on that?

Joe: I do, I I think it would help. Yeah. Yeah. Um, you know, if you, if you think about the way, um, we described prosecution earlier where it's like the, the financial institution does kind of an initial investigation sends their findings to law enforcement. Law enforcement adds their own investigation information and sends to prosecution. Um, you know, there, there's a, we call it signal loss. There's a lot of signal loss in the communication channel. So one of the things we observe at Hummingbird is. You know, our interface, the interface for the, the financial institution, the bank, or the crypto exchange. Um, we try to make as rich as possible with insights and data and the ability to really do high information density displays so that the, the person viewing it can really see what's going on through, um, you know, data visualization and different tools.

The, the report then to law enforcement, sadly, has, has to go through the SAR filing channel where it gets compressed down to a written report, right? So you're losing the richness of some of that data. You're losing the, the visual element of data visualization throughout. Um, and you're trying to compress it as we've discussed down into like what a human can kind of like effectively communicate through written, uh, prose.

Um, I, I think personally that that probably results in more signal loss than we think. And, um, getting, you know, great tools in the hands of law enforcement would ha, you know, probably help us, uh, fight these crimes, uh, even better than they're already doing today. And, uh, I don't mean that as a criticism at all.

I think those, those folks are doing great work. It's just, um, you know, is there, uh, opportunity here for design to help? Yes, probably. Yep.

Stephen: They're doing great work, but they only have so many resources in the day. They only have so much time and, you know, capital to invest in a lot of these things they need to leverage technology. I'm curious, you know, especially on this podcast we've talked about, you know, bad UX usually drives people from more compliant fintechs and crypto exchanges to maybe less compliant ones that have almost like frictionless, seamless interfaces.

Alright. What are your thoughts and do you think, you know, the compliance friction actively creates potential harm to customers because they're, they're going to these easy click places a lot of times where, you know, the friction from compliance is unable to give that same user experience.

Joe: Yeah, you know, you know where I think most, you know, normal folks feel the influence without even realizing it, of, of compliance. It's just in delays to stuff. Um, and I, I've had that personally where, you know, if I made a, a transfer between my accounts or tried to send somebody money for something or, um, move things around.

Those transfers can get delayed, um, by compliance groups that need to review, you know, those transfers. Um, and so I think it actually can be a, a detriment to the business and the customer support that the financial institution provides. Um, you know, a lot of times too, it's funny, Stephen, like we, when we go into a sale to, you know, try and sell Hummingbird to an organization, sometimes we'll take a look at the current setup that the compliance team is working with.

And wow, there are some absolutely brutal, um, you know, tools and setups and things like that out there where the poor compliance people trying to do the job and, and just like spending all day moving data between system A and B and C and manually copy and pasting stuff and dealing with validation issues and stuff like that.

It's like. I'm always, you know, surprised that, you know, those folks are able to continue and, and actually do the work effectively at all.

Stephen: You are surprised that, you know, Google Sheets is probably a core part of most compliance department's investigation technology stack. Unfortunately, um, it's, I

Joe: it was all a

Stephen: into fintechs and cryptos because they're spending so much of their technical expertise on building better products and better solutions that the compliance department is left with whatever leftovers, um, or time that they have, which is usually not very much.

Joe: But I mean, as well, they should, right? Like, I, I don't like, I don't think, uh, um, a compliance, I, I don't think a financial institution should spend its engineering time and engineering dollars on compliance tooling. Like why, why would you? Right. The compliance. Obligations may not be exactly the same as the next financial institution, but they're gonna be similar and they're dictated by the same rule set.

Um, and there's really no like, competitive differentiation that comes from building in house. And so, um, you know, it's, I'm always a little bit mystified when folks wanna put their engineering effort towards, uh, this particular practice area. There. There's just better ways to do it.

Stephen: and I think the, you know, the elephant in the room is that, and never, they never build it for scale. It's just building it to like survive. Uh, so you never get that scaling aspect in those in-house tools. It's usually as like, Hey, we're just trying to hold everything together so it doesn't fall apart, and it's hard to evolve past that point.

I'm curious, as we talk about like ux, you're a project, you have a power background, um, but you're also responsible for running the company as a CEO if you lose a customer. How do you decide if this was like, oh, I, uh, you know, uh, due due to a compliance issue or like a product design failure, like how do you balance, how do you kind of approach the afterthought of like, Hey, where could we have done better?

Is it more finding the financial crime or is it a slicker, seamless dashboard that they could have used a lot more easily in order to find that financial crime?

Joe: Yeah. Uh, well, I, I'll, I will say upfront, we have never lost a customer for a compliance issue. So, um, just to get that out there and be very clear about that. Um, in many ways our platform is a, a massive, you know, level up over, um, what the institution was doing before, and regulators love to see it. Um, no, we, you know, but churn happens, right?

We, we lose customers for, um, you know, a couple of reasons. One is, um, you know, the customer's business model changes and they don't require. A compliance program or they pivot or they go outta business if they're, uh, you know, a FinTech or a startup or something like that. Um, we see those customers churn.

Um, the other is usually a, a cost driver or some sort of material change in the financing of the business where, um, you know, they either decide to go with generic tooling to save money or, um, something changes where, um, you know, cost becomes such a factor that they, they just don't want to invest in the, the program and its software.

So.

Stephen: What, what's interesting, you sparked an idea in my mind where, you know, a lot of our guests have come on and basically said, you know, stable coins succeed despite compliance, not because of it. Do you feel or have an example of where compliance actually unlocked adoption? Adoption instead of slowing it down?

Like, I see the benefits of being highly compliant for a lot of institutions involved with digital assets, but from a project side and from running hummingbird, uh, is there any examples where you're like, no, compliance is making this more attractive, not less attractive.

Joe: I think you see it in, um, you know, onboarding and risk decisioning and underwriting decisions. Um, you know, your, your ability as a lender or a credit provider or something like that, your ability to make those decisions quickly and efficiency can, can be a, uh, a, a major differentiator. Um, I, I'll give the story of, um, square in the early days, um, you know, people think about square's origins, and of course they're correct to think about like the cool factor of, uh, taking payments on your cell phone.

They were, you know, square was an innovator there, but there's a second and very important innovation that Square had, and it's what I just talked about, you know, prior to Square. Um, you know, signing up to be able to take payments, signing up for a merchant account with a, a, you know, a payment processor might have been like a multi-week process, um, for, you know, a food truck vendor or, uh, a retail storefront.

Uh, square's big innovation. There was, you'd sign up online and Square would make the decision and set you up, you know, in that web session in a matter of minutes. Um, so they massively compressed the time to, um, be approved as a merchant acquirer, or not a merchant, acquirer merchant. And, um, you know, that was actually one of the fundamental innovations that I think then led to the just massive growth of that product, um, into all sorts of different payment industries.

So, but that's,

Stephen: still haven't caught on. It still takes three weeks for me. For me, for me to register a business account with the bank, you might as well just give 'em your first born. It's, it's quicker.

Joe: It's a, yeah, I mean, it's, it's, you know what, what's happening behind the scenes is fundamentally a risk decision that the, the, you know, the car, the bank or, or square or whoever is making right, they're saying like, yes, I can trust, you know, this person to take these payments and not commit fraud, um, to a certain degree.

So your ability to make that decision faster, um, can be a massive business driver.

Stephen: I am curious, we've seen so much political unrest just in the last few weeks, since 2026 started. Obviously everything that's happening in Gaza, Russia, Ukraine, Israel, how do we make sure that we're still able to offer these financial services in these high risk areas in jurisdictions and regions, but without de-risking the people that need access?

Probably the stable coins and cryptocurrency even more. I think we saw a chain analysis released a report that, you know, I ran crypto usage was up since last year. Even before what happened, um, recently. How do we make sure that we're, you know, mitigating our risk but not excluding the people that might not be directly involved of to why it's risky.

Joe: Yeah. Um, it, it's absolutely, uh, the quality of your investigation work and the quality of your data, right? I mean, what you need in these circumstances is, um, you know, high quality information about the transactions, who the counterparties are, um, and then vigilant staff, vigilant compliance folks that are making sure you're not violating sanctions obligations, you're not facilitating financing of terrorism.

Um, you know, our, our platform is used in all of these practice areas called CFT, counter Financing of Terrorism. Um, and, um, it really is I think a, a similar investigation practice area, but with a lot more detail and a lot more, more scrutiny and attention to detail. So you just have to be extraordinarily buttoned up about what you're looking out and, um, and the, the risk based judgments that you're making about transactions.

Stephen: You know, thinking back to your data science days is, is how important do you think that is? I know for your platform, the collection of all this data and putting it to an easy to use dashboard is obviously, you know, detrimental for these companies to be able to assess risk fast, like you said, and be able to make these decisions.

But even as a business owner understanding, you know, conversions and close rates and leads, how do you, uh, like where do you think, you know, especially into 2026, how important will data be for your businesses and the industry as a whole?

Joe: Very. Um, and, and you know, I think we've carved out a, a unique spot in the, the data world for Hummingbird, where, um, our primary role, um, as it pertains to machine learning and, and training models is, um, we, we are a, a structured data and labeling provider. So we kind of close the loop for, uh, machine learning training.

Um, the way we do that, if you think about. This is, uh, where we're gonna go. Uh, we're gonna spend a minute going a little bit deeper into crime. If you think about first party fraud, so if I steal your credit card, Stephen, and you report that through a chargeback, you're like, I didn't, you know, I didn't buy those shoes or whatever.

Um, then, you know, we can train based on how many traction, uh, transactions went through and how many later, uh, manifest as chargebacks. And that's a super well labeled training data set for a machine learning exercise. Um, if you're looking for, you know, risk decisioning, um, or, um, fighting anti-money laundering or things like that, there's no, there's no chargeback.

A money launderer is gonna try to avoid your detection at all costs. So there's nothing at the end of that to tell you whether or not the transaction was money laundering. Um, one of the things that the Hummingbird platform does is we create a way to take all of the information that's going into the investigation, gather the insights from the human investigator who is providing some insights in a decision, and create structured and well labeled training data from that data set, deliver it back to our customers, um, and they can train their detection algorithms on that information.

So that is how I think we get better at fighting some of the more complex crimes and the stuff where, um, you know, it's not first party fraud, which is serious of course, but it's more like, um, large scale money laundering to hide organized crime or, or large scale networks.

Stephen: And is your platform. Then training on all the sars they're filing and then giving them even more rich insights. So if I file a SAR on customer A, and you know, six months later, customer B, you know, I'm looking to file a SAR there, and then I see that there's a counterparty transaction between the two of them.

It can like pull all that data together to gimme a whole, a fulsome, more holistic look of my wrist.

Joe: We're more of a data platform, so we're not, uh, we're not, uh, selling anyone algorithms. Uh, our, our role in this game is just data, data, data. We're gonna give you as much clean well labeled data as we can. We're gonna close the loop. We're going to, um, provide you with the tooling you need to run your algorithms as well.

Um, but, uh, those algorithms are, uh, developed by each financial institution. Um, and really the re there's a good reason for that. Like the, the old way of selling software in our business was to. Take a library of like 300 different rules-based algorithms and sell those to financial institutions. Um, guess what, like doesn't actually catch that much crime.

Right? Those rules are pretty basic. They don't actually handle much. Um, the, the modern and advanced companies that we work with, they're developing their own algorithms that are very tailored to their own business. Uh, they're much more aware of the unique risks that they have, um, and, uh, they're much better at detecting stuff and, and blocking it.

Stephen: That makes a lot of sense. You know, I was reading your LinkedIn, you made a post about cybersecurity, keeping you up at night. Funny enough, when I worked at the bank almost 10 years ago, the president of the bank said the exact same thing. Why do you think we haven't solved for this nightmare of breaches, hacks, and other cybersecurity exploits by now?

It's been decades now. We're still scared, sleepless about this scenario. Uh, what do you think the big gap is that we haven't been able to solve for this?

Joe: I think it's a gap. I, I think it's just an ongoing and dynamic situation and the reality of, uh, you know, crime and bad behavior on the web in including either their financial crime or, uh, cybersecurity crime and, uh, hacks and things like that. Um, you know, the reality is like, uh, these types of things can be profitable and that means people are gonna pursue 'em.

And, um, you know, people pursue 'em, the people pursuing 'em are smart and creative and they use the latest technologies and they collaborate and they study, you know, different ways to, you know, perform hacks and things like that. So it just is, um, I think it's the kind of thing where the work is never done right.

Um, it's a inherently, somewhat reactive, uh, part of the business where. You, you just gotta be vigilant and on your toes all the time. And, um, we've really tried to build security into the foundation of our thinking, both in product that we build, but also just team and practices. 'cause social engineering is another major attack vector.

So, um, everything we do, we're thinking about security and how we can always be getting better at it.

Stephen: You're a product guy. What's your favorite features? There might be a couple at high level that you just absolutely love you thought of, or someone brought to you and you're like, oh, this is amazing. What are some of your favorite features of the Hummingbird platform?

Joe: I mean, I, I, I would be remiss not to mention some of the AI utilities that we've built. Um, you know, we, we have done things that are very tailored to our industry and to our mission. And, um, it is harder than you think to get large language models to kind of say the right thing in a compliance environment.

Um, a lot more work than I think people think. And, and it's like, no, you can't just go and fire up a, uh, open AI account and use chat GPT for that. It's not gonna be fit for purpose in the compliance industry. So, um, those are very exciting. Those are a key part of our focus and innovation. Um, I, I'll personally, I'll give you some other things that are not AI as well, like, um, we, we announced our, uh, transaction monitoring service, uh, which one of the things I'm really proud of is in the transaction monitoring business.

There's kind of a way of doing things that's been done for the last 30 years or so, and, um, just like really didn't want to do that. We didn't wanna be the hundredth company to come along with another regurgitation of the same monitoring protocols. Um, and so we took a risky and very differentiated approach and, um, so far it's been about a quarter and a half that we've been in market with it.

Uh, the reception has just been incredible. So, um, I could go into that a little bit if you're interested, but, um,

Stephen: would love to know what the unique approach is as someone that's been doing transaction monitoring for like 10 years. I'm, I'm curious, everything looked a little bit the same, so I'm curious what approach you took.

Joe: I mean, you, you know how it is, like the, you set up with a new transaction monitoring vendor. First step is to like get their API documentation. Second step is to understand their data schema. Third step is to spend several months like transforming your data and sending it into that vendor so that they can then monitor it.

Um, we built an approach that sits natively on top of the cloud data store that you already have. So, snowflake, AWS, something like that. You use your own data schema and rules. You don't have to transform anything, you don't have to send anything to Hummingbird. Um, and we monitor within that environment and then, uh, bring the alerts into Hummingbird, where our, um, incredible case management platform takes over.

Stephen: Yeah. And from a timewise, how much time do you think from like implementation and just all the, all the stuff that the companies had to do in order to get ready to use the transaction mining tool after the salesperson told 'em it was only gonna take a few weeks and they're still waiting three months to implement the transaction mining tool, how much time do you think you're reducing, uh, by laying it on top of an existing platform that they're using?

Joe: there's, there's two parts to that setup, right? One is technical integration and setup just to get the data in the right place and in the right format and things like that. And then development of the monitoring program. Um, development of the monitoring program is the same, right? You, you still have to, um, create the algorithms and things like that.

Uh, our specialty is basically all of the data work and engineering work that goes into the setup is basically zero. I mean, it's not zero, it's like a few minutes or something like that, but it's near zero as you can get versus, uh, what is often a multi-month or a multi-year process to get set up.

Stephen: But then they're also pulling it into a hummingbird. So it's like your case management system's already weighing and you're pulling that information in there where a lot of the transaction monitoring tools I, I've seen, they have a case management system, but it's not necessarily the case management system that the company's using.

'cause the company may have a case management system based on other, you know, in data that they're receiving in about their customers. So I think that's an interesting thing that everything will be housed in Hummingbird in one place.

Joe: Yeah, you can actually send your alerts into another case manager from our transaction monitoring product. Um, we don't, we don't recommend it 'cause there's no other case managers as good, but it can be done.

Stephen: Are you, so are you shocked? We've, you know, are you shocked about how bad the case management system of some of these companies where we talked about, you know, some of the tech stack they, they were using in general, but case management. Are you a little surprised about how, you know, how poor the case management, how, you know, especially with so much cybersecurity risk, why companies aren't spending a little bit more time making sure that they have a secure case management system in place?

Joe: I just knowing how product orgs work, um, you know, the, the reason for that is very simple, which is the, the transaction monitoring product gets all the attention and the engineering and things like that. Um, the case management from legacy vendors in particular. Tends to be an afterthought, sort of a, just like there has to be somewhere to display the alerts generated.

So we'll create a screen that's a list and has some simple workflows. Um, you know, there's a few out there that have, um, I think over the last five years tried to improve, which is great. I think, you know, I, I, I hope that one of our influences on the industry has been to encourage folks to like actually think about the investigator experience.

Like that is an important part of the business, and that was our founding, um, thesis. Um, and we've seen others follow suit, which is great. You know, if we, if we can level that up even across competitors, that's great. Um, but yeah, there's a lot of stuff that's just very manual and very basic, so,

Stephen: you know, rumor has it, speaking about UX and making and up upscaling, you know, what case management looks like. Rumor has it, you spend a lot of time on the company logo and, and the UX design of the dashboard. Uh, how do you approach that? You have a lot of, you have a couple co-founders. How do you approach your love for the design, but also co-founders that might have their own approach to, you know, maybe speed or, you know, other technical aspects of the platform?

Joe: We've always been unified about that. So I've, I've got, uh, three other co-founders, uh, Matt, who I mentioned is a former regulator, so he's kind of our subject matter expert, uh, co-founder. Um, and then Jesse, our CTO and Ryan Gerard, our principal engineer, our both product folks. Uh, they're both engineers. Um, incredibly talented engineers that, um, I worked with at Square.

And, um, one of the things that we learned, uh, working together at Square is just like we love building and shipping product together. We have a shared philosophy that design matters and can be an important differentiator. It shouldn't just be an afterthought. I think the time at Square really taught us that, and we sort of inherently believed that anyway.

Um, and so it's, it's never been an issue. I think we all collectively have always seen and, and believed in the value of investing in design. Um, you know, and I think sometimes people hear the word design and they think aesthetics. Um, aesthetics are obviously important, but we're, what we're talking about when we talk about design is, um, fit for purpose, uh, UIs and user experience.

So the, the product is designed to do the thing that you need to do. It's configured around those use cases and those goals you have. And, um, it's been streamlined to make that as easy as possible for you, it's less about the aesthetics and more

Stephen: yeah, I think that's where I, my head goes when I hear about aesthetic. I've also seen your platform and I'm like, that was the one thing that stuck out in my head. Like, oh, they really thought about how they positioned a lot of the features, the buttons, the kind of workflow. And I think if you don't think about it and you don't come from a background of filing SARS or doing that kind of reporting, it really takes away a lot from the platform if that isn't your primary goal.

So good on you guys for actually focusing on that going forward. Any exciting news coming out of, you know, hummingbird as we're in 2026? Anything exciting or even in the industry that you're seeing that you're excited about?

Joe: Um, I'm, I'm excited about a lot. I mean, um, I think one thing that's really cool is in, in investigation work is we're seeing a lot more application of automation and AI in ways, um, that can basically eliminate the need for human labor on the really tedious and, and, um, low value stuff. Allow that, that, you know, the humans in the equation to go and focus on some of the more complex crimes.

Um, the, you know, the, the pity about the industry that happens today is kind of like some of the more complex stuff. Probably doesn't get the investigation attention it needs because, you know, the org is being forced to just move through procedural work to comply with, with compliance. And so there's some sort of perverse incentive to like, do the minimum possible to check the box in compliance versus like really pay attention to complex investigations.

And, um, our customers love that they can kind of shift that a little bit and automate, um, you know, the stuff that would've been just low value work and, and refocus on stuff that actually matters. Um, so that's really exciting. I'm excited too. You know, I've been in the, in or around the crypto space for a long time now, almost 12 years.

Um, I think the, you know, finally seeing some federal acceptance of crypto and, and the ability to, um, have, have regulations that keep the ecosystem safe and consistent and allow the big banks to kind of dive in, but also like contribute, safeguard measures and, and diligence and things like that. I think that levels up the overall safety and the overall crime fighting potential of the whole ecosystem.

Um, so I'm, I'm bullish about that effect as well.

Stephen: I love it. Have you guys ever thought about the other side of the table? Like you're creating something that gives investigators an opportunity to not have to focus so much on writing quality, SARS and str. How about the regulators that are receiving it? You know, is there any way that you can help synthesize that on the other end, where they're able now to compile all that data and figure out what's more urgent?

Which sars more related to what's happening in the world? Because I think that's the downfall is we're, we're filing all the sars, we're making good quality, but they don't have enough time on the other end of the resources to kind of compile it and make sure to trigger what's most important. Any thoughts about helping the regulators on the other side, uh, with the data?

Joe: Regulators and law enforcement. Yeah. Yeah, absolutely. I think, um, that's, uh, probably a future market for us. It's one that I'm excited about there. There's a lot of help needed over there and a lot of good work happening that, um, we could, we could assist with. So.

Stephen: like I can picture with your filing like with your platform, almost like real time sars getting into law enforcement hands. That would be, that's probably the utopia where investigators can actually see the fruits of their work be actioned right away, or at least within months versus maybe decades. Uh, as A-C-E-O-I think you invest in companies on the side as well.

You were early on in Circle and some other companies that iPod. I know we talked earlier before we recorded about, you know, spending some time doing what you love doing, which is music. What are some other ways maybe you, you know, step away from your job to make sure you're fresh and relaxed and concerning some of your energy for things that you love to do.

Joe: Love music. Um, I'm actually recording this in a podcast studio at this coworking space I found with a music studio kind of built into it, which is really cool. Um, I love food and, and restaurants and things like that. Um, I'm super into, um, fitness and, and working out and, uh, I love to go. Um, rucking, have you ever heard of that?

It's like you put on a heavy backpack and go walk around. Those are some of my hobbies.

Stephen: people is heavy. My wife got one that's 12 pounds. I'm like, oh, 12 pounds. Like I almost died the first time. I think people don't realize like how, I'm like, yeah, I'll probably take a 50 pound pack. I don't think people realize how heavy things are. Even when you lose maybe 50 pounds, you're like, oh, it should be easy just to throw 50 pounds on my bag.

It's not as easy as it looks. I have to say. Maybe I'm just soft.

Joe: it's a, no. It's a great exercise. I love it. It's, I, I used to be a jogger, which, you know, jogging is cool, but I, I found it, it was kind of a little hard on the joints and, um. I, I now go like hiking and, and go rocking with this, uh, weighted backpack on. And I've been doing that for about two years now, regularly, and I just love it, so.

Stephen: What's, what's the weight you're carrying around? So I have something to aspire to.

Joe: Well you do, I I will say, like you, you're exactly correct. It's best if folks start low and kind of work up, um, because it is, uh, it is a different thing on your shoulders and your back. Um, I'm doing about 45 pounds now,

Stephen: Oh, wow. Oh, wow. All right. I got a lot of work to do. Uh, for you as a CEO. We, we have a lot of CEOs, founders that listen to this podcast. Is there any books, podcasts, ideas that help shape your thinking? You were able to pick the, you know, some of the best companies in the world very early on, so it's almost like you're seeing around corners or just some intuition to join.

These companies are gonna be changing the world, including building hummingbird. Uh, what are your thoughts? Anything that stands out that inspired you or helped you with some of your decision making?

Joe: Yeah. Yeah. Um, well, and, and, um, just I think for your, your listeners who are interested in finance and the future of finance and, and technology and stuff like that. Um, you know, a couple of recommendations like one is, um, there's a guy, he is a vc, Morgan Housel. Um, he's written I think three books. I've read two of them on basically, um, I think his first book is called The Psychology of Money.

The second book is called The Art of Spending. Um, both are not finance books in the sense of being like tactics or how to guides or market commentary. Um, his sort of unique, you know, perspective on the world is the emotional relationship and the psychology and things like that, that we have with our money and with money generally, and how we perceive wealth.

Um, and I would recommend those to, to anyone. I think those are great books and a really interesting perspective. Um, I read, uh, I listened to, um, a book called Die With Zero that was, uh, profoundly influenced me last year. Um, you know, it's, it's the idea that, you know, you should go. I, I'm, I'm personally, I'm biased towards, like, I don't like to spend money.

I tend to like, kind of try to save as much as possible and don't love to spend on things and agonize over

Stephen: the other books a couple times to get to maybe.

Joe: And, you know, di Die was Zero was I think around, you know, some concepts that originate in, in Buddhism and, and it, it's like the trying to experience the joy of giving and of using your, your resources to improve your circumstances and those of your loved ones and things like that.

And. I just, uh, it was a new perspective for me and one that I was like, it was super thought provoking. Um, and then podcasts, uh, this podcast of course is great. Um, I love, uh, I love the Waking Up app, which is a meditation app done by Sam Harris. And, um, it features a lot of podcasts actually on, um, neuroscience, psychology, um, meditation, um, you know, and, and that, that just in my opinion, is one of the best resources out there for that kind of thing.

So if you're into that, that's another good one.

Stephen: I love it. And it's funny, I was just listening to the, my First Million podcasts the last few weeks, and they mentioned both those books as they have, you know, some of the most successful founders. So it seems like, you know, your successful CEOs have a little book club going around that nobody else knows about, where you guys, you guys have a little book list that you have.

Uh.

Joe: yeah, they're not, they're not tactical books in any stretch of the imagination, but I, I thought it was a really fresh look at like, our cultural obsession with money and wealth and, um, where that's good and where it's bad and you know, that kind of thing.

Stephen: That's amazing Joe, where's the best place people can find? We'll definitely link Hummingbird and tell people to check out a demo. 'cause it is pretty cool if you work in the crypto payment space, FinTech space, you definitely have compliance requirements and this is probably gonna be the one that you see the most value in my opinion as an investigator.

You know, spending 35, 40% of your time doing the procedural work that you say, the low, the low value work, uh, isn't the best, you know, for any of these companies. So where's the best place people can find more of your thoughts? Is it LinkedIn? I know they got rid of crypto Twitter. I think they're in the midst of getting rid of crypto Twitter, so I'm not sure if you're, you're salvaging your last few weeks there.

Joe: I, I have, uh, almost fully extracted myself from social media. Uh, I do not have any social media on my phone or anything, but I do post on LinkedIn, uh, mainly to share updates about, uh, hummingbird. So, uh, that is the best place or just hummingbird.co our website.

Stephen: I appreciate it. Joe Robinson, thank you so much for joining us today.

Joe: Thanks so much. This was fun.

‍