The Connector Podcast - FinanceX #24 - Summer Edition - Fintech’s Trust Reset

WEBVTT
00:00:00.160 --> 00:00:06.639
So back in 2025, venture capitalists poured this uh really staggering amount of money.
00:00:06.799 --> 00:00:12.560
I mean$116 billion into financial technology companies globally.
00:00:12.880 --> 00:00:14.800
Yeah, which is just an absolute mountain of cash.
00:00:15.119 --> 00:00:15.279
Right.
00:00:15.359 --> 00:00:15.759
It is.
00:00:15.919 --> 00:00:22.239
But the wild part is despite all that money, the actual number of deals that got made hit an eight-year low.
00:00:22.480 --> 00:00:22.800
Wow.
00:00:23.039 --> 00:00:23.359
Yeah.
00:00:23.519 --> 00:00:30.399
So the money is clearly there, but you know, the patience for those old fake it till you make it startups, that is completely dead.
00:00:30.719 --> 00:00:33.520
It's uh it's a phenomenal contradiction when you look at it on paper.
00:00:33.600 --> 00:00:37.759
You've got this massive liquidity, but just absolute zero tolerance for hype.
00:00:38.079 --> 00:00:38.399
Exactly.
00:00:38.479 --> 00:00:40.320
And that is exactly why we're here today.
00:00:40.399 --> 00:00:42.000
So welcome to this deep dive.
00:00:42.240 --> 00:00:49.359
Today we're exploring a really fascinating stack of articles and research from the summer 2026 edition of Finance X magazine.
00:00:49.520 --> 00:00:54.399
And our mission for you today is to decode what industry insiders are currently calling the summer compression.
00:00:54.719 --> 00:00:55.600
The summer compression, yeah.
00:00:55.759 --> 00:00:59.200
Yeah, it's this massive, quiet rewiring of the global financial system.
00:00:59.359 --> 00:01:05.280
The flashy parties are over, the noise has vanished, and this rigorous operational discipline is taking over.
00:01:05.680 --> 00:01:10.319
We really are witnessing a profound shift in the architecture of the industry.
00:01:10.480 --> 00:01:19.120
It's a transition from a landscape driven almost entirely by, you know, experimentation to one that is strictly governed by execution.
00:01:19.519 --> 00:01:19.760
Okay.
00:01:20.000 --> 00:01:20.959
Let's unpack this.
00:01:21.120 --> 00:01:26.159
Because Andrew Crin from Video Global, he lays out this core thesis in the magazine perfectly.
00:01:26.400 --> 00:01:33.359
He argues that the first half of 2026 wasn't marked by like a single crazy black swan event.
00:01:33.519 --> 00:01:33.760
Right.
00:01:34.000 --> 00:01:35.840
Or even one sweeping law.
00:01:36.000 --> 00:01:37.760
It was more of a tightening of the atmosphere.
00:01:37.840 --> 00:01:41.120
Like AI is no longer a fun toy you use to write emails, right?
00:01:41.280 --> 00:01:42.560
It's a strict governance issue.
00:01:42.799 --> 00:01:43.040
Absolutely.
00:01:43.280 --> 00:01:46.959
And instant payments aren't some neat infrastructure project to brag about anymore.
00:01:47.040 --> 00:01:50.480
They are a basic expectation, like the lights turning on when you flip a switch.
00:01:50.560 --> 00:01:52.799
It really feels like an industry-wide coming of age.
00:01:53.040 --> 00:01:53.680
Yeah, it really does.
00:01:54.000 --> 00:02:04.799
If the fintech sector used to be uh that teenager who threw wild expensive house parties with their parents' money, this summer compression is that same teenager finally growing up, putting on a suit, and trying to get a mortgage.
00:02:04.959 --> 00:02:07.120
You know, you have to prove institutional readiness.
00:02:07.519 --> 00:02:13.520
I love that analogy because I mean, getting a mortgage requires paperwork, audits, a proven track record.
00:02:13.840 --> 00:02:14.879
Things a teenager hates.
00:02:15.120 --> 00:02:15.520
Exactly.
00:02:15.759 --> 00:02:17.199
Things a teenager hates.
00:02:17.439 --> 00:02:27.680
And if we connect this to the bigger picture, this shift toward maturity is largely being forced by incredibly strict regulatory timelines that are no longer theoretical.
00:02:27.759 --> 00:02:28.400
They have arrived.
00:02:28.719 --> 00:02:29.360
Right, they're here.
00:02:29.680 --> 00:02:32.000
Europe is really the epicenter forcing this right now.
00:02:32.159 --> 00:02:38.080
The EU AI Act becomes a live, enforceable issue on August 2nd, 2026.
00:02:38.159 --> 00:02:39.520
Aaron Powell Wow, so right around the corner.
00:02:39.840 --> 00:02:39.919
Yeah.
00:02:40.080 --> 00:02:49.120
So if you're a financial institution using an AI model to approve credit or uh assess risk, you can no longer just say, well, the algorithm figured it out.
00:02:49.360 --> 00:02:51.199
You have to prove exactly how it works to a regulator.
00:02:51.360 --> 00:02:51.520
Trevor Burrus, Jr.
00:02:51.599 --> 00:02:52.800
And it's not just AI, right?
00:02:52.879 --> 00:02:56.240
I mean, the actual movement of money is getting regulated into maturity, too.
00:02:56.319 --> 00:02:56.960
Aaron Powell Completely.
00:02:57.199 --> 00:03:04.560
The European Central Bank's instant payments timetable made verification of pay mandatory in Euro area countries late last year.
00:03:04.639 --> 00:03:05.039
Aaron Powell Okay.
00:03:05.360 --> 00:03:17.039
And for those who might be unfamiliar, verification of payee just means the banking system automatically checks that the name of the person you're sending money to actually matches the I-BAN or the account number.
00:03:17.120 --> 00:03:17.680
Aaron Powell Right.
00:03:17.840 --> 00:03:19.360
Before the money even leaves your account.
00:03:19.680 --> 00:03:20.159
Exactly.
00:03:20.319 --> 00:03:27.360
Which sounds basic, but implementing that across thousands of banks takes massive engineering discipline.
00:03:27.680 --> 00:03:28.639
Because it's huge undertaking.
00:03:28.800 --> 00:03:28.960
Yeah.
00:03:29.199 --> 00:03:32.879
The old race was simply about who could launch the fastest app.
00:03:33.120 --> 00:03:37.759
The new battleground is entirely about who can be trusted to operate at high speed.
00:03:39.120 --> 00:03:41.599
Trust, not speed, is the new currency.
00:03:41.840 --> 00:03:45.919
So what does this demand for commercial discipline actually look like out in the wild?
00:03:46.240 --> 00:03:50.319
I want to pivot to a specific microcosm from our sources, which is Romania.
00:03:50.800 --> 00:03:51.599
Ah, yes.
00:03:51.759 --> 00:03:52.879
The Keysfin report.
00:03:53.120 --> 00:03:53.199
Right.
00:03:53.280 --> 00:04:01.520
The Keysfin report in our stack looks at the Romanian fintech ecosystem and it illustrates a market that has aggressively made this transition from height to discipline.
00:04:01.680 --> 00:04:04.879
According to the report, Romania is no longer a future or emerging market.
00:04:04.960 --> 00:04:06.879
It is very much a now market.
00:04:07.280 --> 00:04:08.319
The numbers there are pretty wild.
00:04:08.639 --> 00:04:08.800
Yeah.
00:04:08.960 --> 00:04:17.040
Turnover hit a record 82.7 million euros in 2024, which is a 143% jump since 2020.
00:04:17.439 --> 00:04:23.839
That kind of sustained growth is impressive, but you know, the structure of that market is what really reveals the reality of the summer compression.
00:04:24.000 --> 00:04:29.120
Aaron Powell Because the wealth is highly concentrated, five companies control 86% of the revenue.
00:04:29.360 --> 00:04:29.519
Yes.
00:04:29.759 --> 00:04:38.000
It's led by a company called SelfPay, which generated 47.5 million euros in turnover and pulled in 6.5 million in pure profit.
00:04:38.319 --> 00:04:41.040
But I have to push back on this razy picture for a second.
00:04:41.279 --> 00:04:41.439
Okay.
00:04:41.839 --> 00:04:46.160
Because there is a deeply sobering statistic buried in this text.
00:04:46.319 --> 00:04:51.279
Out of 29 companies analyzed in that report, only 12 were actually profitable last year.
00:04:51.519 --> 00:04:51.600
Right.
00:04:51.839 --> 00:04:56.480
So if over half the market is still losing money, aren't we just looking at a financial hunger games?
00:04:56.560 --> 00:05:00.720
Like eventually the massive traditional banks are just going to eat the survivors, right?
00:05:01.199 --> 00:05:06.399
Well, this raises an important question, and it is a very common fear when an industry consolidates like this.
00:05:06.560 --> 00:05:09.439
But what we're actually observing is a shakeout phase, not a slaughter.
00:05:11.519 --> 00:05:11.680
Yeah.
00:05:11.920 --> 00:05:13.120
The dynamic has shifted.
00:05:13.199 --> 00:05:17.199
It's no longer this zero-sum war between scrappy startups and legacy banks.
00:05:17.680 --> 00:05:21.040
Look at the recent RofinTech Awards Gala mentioned in the source.
00:05:21.199 --> 00:05:25.439
The leadership award went to Omer Tetic, the CEO of Banca Transylvania.
00:05:25.680 --> 00:05:26.639
Which is fascinating.
00:05:26.800 --> 00:05:31.120
A traditional bank CEO winning the top prize at a disruptive fintech gala.
00:05:31.439 --> 00:05:34.240
It speaks volumes about the new consensus, which is a hybrid model.
00:05:34.480 --> 00:05:34.720
Yeah.
00:05:34.959 --> 00:05:39.360
Banks provide the massive scale, the deep regulatory capital, and the consumer trust.
00:05:39.600 --> 00:05:39.839
Right.
00:05:40.079 --> 00:05:45.360
And the highly efficient fintechs, the ones actually surviving the shakeout, they provide the plumbing.
00:05:45.519 --> 00:05:47.839
They build the back-end infrastructure that the banks use.
00:05:48.240 --> 00:05:49.120
Making things run smoothly.
00:05:49.439 --> 00:05:49.839
Exactly.
00:05:50.240 --> 00:05:58.319
And the margins these plumbing companies are achieving are staggering because they aren't burning millions of dollars on consumer marketing or customer acquisition anymore.
00:05:58.560 --> 00:06:02.879
The report highlights a company called FinWare with a 31.8% net margin.
00:06:03.120 --> 00:06:03.600
That's huge.
00:06:03.920 --> 00:06:09.040
And Bull Technologies, which reportedly operated with a 148.9% EBITDA margin.
00:06:09.360 --> 00:06:09.759
Wait, hold on.
00:06:09.839 --> 00:06:10.560
Let me stop you there.
00:06:10.800 --> 00:06:13.759
148.9% EBITDA margin.
00:06:13.920 --> 00:06:23.680
I mean, I might just be a podcaster, but mathematically, you cannot have a margin over 100% unless there is some serious accounting sorcery going on.
00:06:23.920 --> 00:06:25.680
Margin is a percentage of revenue.
00:06:25.839 --> 00:06:27.279
You can't keep more than you make.
00:06:27.680 --> 00:06:30.000
You caught that, and you are entirely correct.
00:06:30.160 --> 00:06:34.000
It is mathematically impossible in standard day-to-day operations.
00:06:34.319 --> 00:06:35.199
Okay, so what's going on there?
00:06:35.439 --> 00:06:43.600
What the Keysfin report is likely capturing there with bold technologies is a specific accounting anomaly for that fiscal year.
00:06:43.839 --> 00:06:49.439
For instance, they might have had a massive one-off asset sale, like selling a piece of intellectual property.
00:06:49.600 --> 00:06:50.160
Ah, okay.
00:06:50.399 --> 00:06:57.600
Or they might be aggressively capitalizing their software development costs, which temporarily inflates EBITDA on paper.
00:06:57.920 --> 00:07:05.040
But even if you strip away the accounting quirks to find their normalized margin, the underlying story remains incredibly strong.
00:07:05.360 --> 00:07:06.800
They're still making a ton of money.
00:07:07.040 --> 00:07:07.279
Right.
00:07:07.439 --> 00:07:13.519
These B2B infrastructure companies are generating a staggering amount of cash relative to their operational footprint.
00:07:13.680 --> 00:07:15.519
They escaped the cash burn trap.
00:07:15.680 --> 00:07:18.560
They aren't trying to be banks, they're just making the banks better.
00:07:18.800 --> 00:07:21.279
They survived the teenager phase and got the mortgage.
00:07:21.600 --> 00:07:22.319
Precisely that.
00:07:22.639 --> 00:07:33.040
So if efficiency and B2B utility are the keys to surviving this compression as a back-end provider, how does a modern consumer-facing digital bank actually survive?
00:07:33.279 --> 00:07:40.800
That brings us to Bunk, the European Neobank profiled in the magazine, because they aren't hiding in the back end, they're out there in the consumer trenches.
00:07:41.120 --> 00:07:45.279
And their blueprint is a masterclass in this new era of operational discipline.
00:07:45.600 --> 00:07:48.720
They are currently in their third consecutive year of profitability.
00:07:48.879 --> 00:07:57.600
They posted 34.6 million euros in pre-tax profit in 2025, serving 22 million users across 30 European markets.
00:07:57.920 --> 00:07:58.480
It's massive.
00:07:58.800 --> 00:08:00.480
And they are expanding furiously.
00:08:00.639 --> 00:08:01.680
They applied for a U.S.
00:08:01.759 --> 00:08:03.199
broker dealer license.
00:08:03.279 --> 00:08:06.079
They filed for a full OCC banking license in the U.S.
00:08:06.160 --> 00:08:10.720
in January 2026, and a Mexican banking license in May 2026.
00:08:10.800 --> 00:08:14.720
Aaron Powell I want to pause on that OCC license for a moment because it really highlights their ambition.
00:08:14.879 --> 00:08:24.079
The Office of the Controller of the Currency in the U.S., getting an OCC charter means you aren't just partnering with a small regional bank to borrow their regulatory umbrella, which is what most startups do.
00:08:24.160 --> 00:08:24.560
Aaron Powell Right.
00:08:24.639 --> 00:08:26.240
The usual banking as a service play.
00:08:26.639 --> 00:08:27.120
Exactly.
00:08:27.279 --> 00:08:32.799
You are applying to be a fully regulated, direct national bank in the United States.
00:08:32.960 --> 00:08:35.600
It is a grueling expensive process.
00:08:35.919 --> 00:08:44.000
And to handle that kind of massive global footprint while staying profitable, the article notes they run heavily on an AI assistant named Finn.
00:08:44.240 --> 00:08:45.600
Aaron Powell Yeah, Finn is fascinating.
00:08:45.919 --> 00:08:47.919
It's not just a chatbot for customers.
00:08:48.159 --> 00:08:56.000
Finn is woven into their actual back office operations, automating fraud detections, support routing, and compliance checks.
00:08:56.159 --> 00:08:58.240
That keeps their headcount lean while they scale.
00:08:58.559 --> 00:09:03.200
They are arguably one of the busiest neobanks in the world from a regulatory bandwidth perspective.
00:09:03.279 --> 00:09:05.120
Aaron Powell But here's where it gets really interesting.
00:09:05.279 --> 00:09:08.799
We have an interview in the stack with their chief evangelist Joe Wilson.
00:09:09.039 --> 00:09:12.639
The inviewer practically begs him to talk about their massive competitors.
00:09:13.200 --> 00:09:17.120
Companies like Revolute, which absolutely dominates places like Ireland and much of the UK.
00:09:17.360 --> 00:09:17.519
Right.
00:09:17.759 --> 00:09:20.000
And Wilson flat out refuses to engage.
00:09:20.159 --> 00:09:24.000
He says, you can't build something extraordinary by following someone else's stuff.
00:09:24.159 --> 00:09:25.519
You'll always be chasing.
00:09:25.759 --> 00:09:27.519
I have to challenge this logic, though.
00:09:27.679 --> 00:09:37.279
Is it really a sustainable strategy for a global bank to just put on blinders, ignore trillion-dollar giants in your sector, and pretend you're operating in a vacuum?
00:09:37.679 --> 00:09:41.440
It sounds counterintuitive, perhaps even a bit arrogant from the outside.
00:09:41.600 --> 00:09:46.159
But if you synthesize Wilson's broader philosophy, it's actually hyper-pragmatic.
00:09:46.879 --> 00:09:51.919
Bunk isn't trying to be the universal everything app for every single person on Earth.
00:09:52.240 --> 00:09:52.399
Okay.
00:09:52.720 --> 00:10:01.519
They are obsessively focused on one very specific demographic, digital nomads, expats, and internationally minded professionals.
00:10:01.919 --> 00:10:05.120
The people whose lives don't fit neatly inside one country's borders.
00:10:05.519 --> 00:10:06.320
That's the core of it.
00:10:06.559 --> 00:10:10.639
Wilson estimates there are about 50 million people in this specific category right now.
00:10:10.799 --> 00:10:16.159
So Bunk's entire product roadmap isn't driven by what Revolute just launched in London.
00:10:16.320 --> 00:10:20.960
It is driven entirely by their own audience's need for frictionless cross-border living.
00:10:21.279 --> 00:10:22.000
That makes a lot of sense.
00:10:22.320 --> 00:10:25.679
These people need an instant, fully functioning bank account the morning they land in a new country.
00:10:25.840 --> 00:10:31.759
They need seamless currency handling and travel insurance that actually follows them across borders without fine print exemptions.
00:10:32.000 --> 00:10:32.720
Okay, yeah.
00:10:32.879 --> 00:10:41.919
If your specific user base tells you they are all taking remote jobs in Mexico, you don't look at a competitor spreadsheet, you just go file for a Mexican banking license.
00:10:42.240 --> 00:10:45.360
Their strategy works because the audience dictates the engineering.
00:10:45.679 --> 00:10:56.879
But to serve that audience of 50 million global nomads and to power those highly efficient new ecosystems we saw in Romania, the literal pipes moving global money have to be upgraded.
00:10:57.200 --> 00:10:57.360
Absolutely.
00:10:57.679 --> 00:11:05.039
If bunk is securing all these licenses to cater to nomads, they can't rely on the old fragmented banking pipes to move that money.
00:11:05.200 --> 00:11:08.960
They need infrastructure that crosses borders as easily as their customers do.
00:11:09.039 --> 00:11:17.279
Aaron Powell And historically, crossing borders is where financial plumbing gets incredibly expensive, painfully slow, and highly prone to errors.
00:11:17.360 --> 00:11:17.840
Aaron Powell Right.
00:11:17.919 --> 00:11:23.120
So according to the magazine, there is a new solution from the European Payments Council, or EPC.
00:11:23.279 --> 00:11:26.080
It's called the One Legout Instant Credit Transfer Scheme.
00:11:26.159 --> 00:11:28.720
Aaron Powell Thankfully they shortened that to OCTINST.
00:11:28.960 --> 00:11:30.639
Yeah, OCTINST is much better.
00:11:30.799 --> 00:11:39.519
Basically, it connects the blazing fast European SEPARails, the single euro payments area, which handles domestic European transfers with non-Euro global systems.
00:11:39.759 --> 00:11:43.039
It allows cross-border payments to happen in seconds, not days.
00:11:43.360 --> 00:11:44.000
Huge upgrade.
00:11:44.399 --> 00:11:48.799
And it does this using a rich data standard called ISO 2022.
00:11:49.200 --> 00:11:53.759
I look at this and think of it like a universal travel adapter for global money.
00:11:54.080 --> 00:11:55.840
That is a brilliant way to frame it.
00:11:56.000 --> 00:11:58.559
Because before this, the plug simply didn't match.
00:11:58.960 --> 00:12:01.840
But let me push back on the necessity of this for a second.
00:12:02.240 --> 00:12:10.399
We have massive tier one banks out there, institutions with trillions of dollars under management that have been moving money globally for a century.
00:12:10.720 --> 00:12:17.679
Why is a wonky regulatory scheme like Ofi TINST so crucial for the smaller players and neobanks we've been talking about?
00:12:17.840 --> 00:12:20.240
Didn't the big banks already solve cross-border money?
00:12:20.559 --> 00:12:27.039
Well, they solved it, but they solved it through a very closed, expensive system known as correspondent banking networks.
00:12:27.279 --> 00:12:32.480
Imagine a customer trying to send money from a local bank in Spain to a vendor in Singapore.
00:12:32.799 --> 00:12:40.799
Under the old system, your money didn't just fly directly there, it had to bounce through three or four different intermediary banks across the globe.
00:12:41.120 --> 00:12:42.159
Like connecting flights.
00:12:42.559 --> 00:12:43.039
Exactly.
00:12:43.279 --> 00:12:46.799
And every single one of those correspondent banks acts like a toll booth.
00:12:46.960 --> 00:12:52.320
They take a fee, they do their own compliance checks, and they add days to the transaction time.
00:12:52.480 --> 00:13:00.240
Only the absolute largest monolithic banks could afford to build and maintain the liquidity for those massive global networks.
00:13:00.320 --> 00:13:05.919
Aaron Powell So smaller players like the innovative startups were just priced out of offering a competitive cross-border service.
00:13:06.240 --> 00:13:08.080
They were effectively locked out of the global game.
00:13:08.399 --> 00:13:13.840
But OCT Inst changes the paradigm because it is evolutionary, not disruptive.
00:13:14.080 --> 00:13:21.600
It allows a smaller payment service provider to take the instant domestic infrastructure they already built and simply plug it into this global adapter.
00:13:21.919 --> 00:13:22.559
Oh, that's clever.
00:13:22.879 --> 00:13:25.360
Plus, you mentioned ISO 2022.
00:13:25.600 --> 00:13:32.480
That standard allows incredibly rich data, like full invoice details, complex names, and addresses to travel with the money.
00:13:32.639 --> 00:13:38.159
It stops the transfer from getting held up by an automated compliance flag in another country just because an initial was missing.
00:13:38.320 --> 00:13:38.960
Oh, that makes sense.
00:13:39.279 --> 00:13:48.879
The source even cites the Spanish market, where local payment providers found that implementing OCT inst was actually cheaper than setting up their initial domestic instant payment systems.
00:13:49.120 --> 00:13:49.759
Wow.
00:13:50.080 --> 00:13:56.080
That is a massive leveling of the playing field because customers don't see borders anymore, they just expect speed.
00:13:56.320 --> 00:14:04.320
Whether you're sending money to a friend across the street or paying a contractor across the Atlantic, you expect the notification to pop up on their phone in two seconds.
00:14:04.639 --> 00:14:11.600
And that very expectation of instantaneous speed introduces the darkest, most challenging aspect of this entire evolution.
00:14:11.919 --> 00:14:12.559
The speed trap.
00:14:12.799 --> 00:14:12.960
Yes.
00:14:13.200 --> 00:14:24.480
If OCT Inst and systems like FedNow in the United States allow global money to move in under two seconds, that is fantastic for Bunks digital nomad buying a coffee in Mexico.
00:14:24.639 --> 00:14:28.559
But it is an absolute nightmare for law enforcement and stopping financial crime.
00:14:29.039 --> 00:14:31.279
Because the criminals are extremely early adopters.
00:14:31.519 --> 00:14:33.919
They get to use those exact same two-second rails.
00:14:34.320 --> 00:14:34.799
Right.
00:14:35.039 --> 00:14:40.879
The source material features a piece from Zovin AI that lays this threat out terrifyingly well.
00:14:41.120 --> 00:14:54.480
If a syndicate hacks an account or runs a scam, they can now use instant rails to rotate those stolen funds across dozens of international accounts, converting it into different currencies before a human compliance analyst has even finished their morning coffee.
00:14:54.799 --> 00:14:55.440
It's just too fast.
00:15:10.159 --> 00:15:14.559
Analysts are just burning out looking at false alarms while the real criminals move at the speed of light.
00:15:23.039 --> 00:15:24.559
But the fraudsters are dynamic.
00:15:24.639 --> 00:15:26.879
They just start sending$9,999.
00:15:27.120 --> 00:15:31.759
Aaron Powell So the proposed solution in the magazine is a leap forward called agentic AI.
00:15:31.919 --> 00:15:40.399
The piece explains that unlike basic machine learning that just spits out a numerical risk score for a single isolated transaction, agentic AI actually reasons.
00:15:40.480 --> 00:15:42.480
Aaron Powell It thinks about the broader context.
00:15:42.559 --> 00:15:42.960
Trevor Burrus Yeah.
00:15:43.039 --> 00:15:53.279
It builds full relationship maps, it correlates the IP address of the sender, the corporate registry history of the beneficiary, the velocity of the transactions, and prior alerts.
00:15:53.519 --> 00:16:00.240
It actually drafts a full, readable, suspicious activity narrative before a human even opens the file.
00:16:00.480 --> 00:16:04.879
By doing that, it drops those false positives by 60 to 70 percent.
00:16:05.200 --> 00:16:07.440
It is a monumental shift in capability.
00:16:07.600 --> 00:16:21.600
It's the difference between a legacy system flashing a red light and saying, this number looks bad, versus an agentic AI saying, here is the entire criminal network mapped out, here is how the shell companies connect, and here's exactly why I flagged it.
00:16:21.919 --> 00:16:23.759
I compare it to upgrading your home security.
00:16:23.919 --> 00:16:26.639
The old rules-based system is like a jumpy guard dog.
00:16:26.720 --> 00:16:30.399
It barks at the wind, it barks at the mailman, it barks at a plastic bag tumbling down the street.
00:16:30.799 --> 00:16:33.120
It alerts you to everything until you just start ignoring it.
00:16:33.279 --> 00:16:33.440
Right.
00:16:33.679 --> 00:16:45.440
The new agentic AI is like hiring a seasoned veteran detective who sits quietly in the corner and only taps you on the shoulder when they hand you a fully researched, cross-referenced dossier of an actual crime in progress.
00:16:45.840 --> 00:16:47.039
A very apt comparison.
00:16:47.440 --> 00:16:56.000
But, and I have to pump the brakes here, if we let an AI reason and act autonomously to hunt criminals, aren't regulators gonna completely freak out?
00:16:56.159 --> 00:16:58.399
We talked about the EU AI Act earlier.
00:16:58.639 --> 00:17:01.519
How do you audit a detective whose brain is made of math?
00:17:01.600 --> 00:17:02.720
Aaron Powell That's the big fear, isn't it?
00:17:02.960 --> 00:17:03.120
Yeah.
00:17:03.360 --> 00:17:09.839
And honestly, if the AI is that fast and that good, the human is just gonna blindly hit approve to keep up with their quota.
00:17:10.000 --> 00:17:11.759
It just turns into rubber stamping.
00:17:12.000 --> 00:17:14.880
How does the industry prevent that black box nightmare?
00:17:14.960 --> 00:17:20.400
Aaron Ross Powell Well, what's fascinating here is how the industry is trying to solve exactly that fear of the rubber stamp.
00:17:20.640 --> 00:17:26.640
There's an excellent article in the stack by Maxine Grindovich from BeckerTech about building intelligent systems.
00:17:26.799 --> 00:17:29.839
The key phrase he introduces is human plus agent.
00:17:30.160 --> 00:17:30.880
Human plus agent.
00:17:31.039 --> 00:17:31.200
Okay.
00:17:31.519 --> 00:17:36.400
The future they're building is not fully autonomous AI running wild in the financial system.
00:17:36.559 --> 00:17:41.359
The AI does the heavy lifting, the immense data triage, the truss referencing.
00:17:41.599 --> 00:17:47.039
But the human being retains the final override, and most importantly, the legal accountability.
00:17:47.440 --> 00:17:52.240
So the UI, the actual software interface, has to force the human to verify the work.
00:17:52.720 --> 00:17:53.200
Yes.
00:17:53.519 --> 00:17:56.079
The source notes the math on the time savings here.
00:17:56.400 --> 00:18:04.000
In a traditional compliance environment, a human investigator spends only 15% of their time applying actual analytical judgment.
00:18:04.240 --> 00:18:12.240
The other 85% is just mind-numbing data, gathering, clicking through different screens, copying and pasting IP addresses, checking corporate registries.
00:18:12.640 --> 00:18:13.519
Sounds exhausting.
00:18:13.759 --> 00:18:14.079
It is.
00:18:14.160 --> 00:18:17.920
In an agentic AI environment, that judgment time shifts to 70%.
00:18:18.720 --> 00:18:24.559
But as Grinovich wisely notes in his piece, complexity loves to wear the mask of sophistication.
00:18:24.880 --> 00:18:26.240
Ooh, that is a great quote.
00:18:26.400 --> 00:18:28.640
What does he mean by it in the context of compliance?
00:18:29.039 --> 00:18:37.519
He means that if you just bolt a highly complex AI onto a messy, disorganized corporate structure, you're going to get an absolute disaster when you try to scale it.
00:18:37.599 --> 00:18:39.039
You will get the rubber stamping you fear.
00:18:39.279 --> 00:18:39.599
Oh, I see.
00:18:39.839 --> 00:18:44.079
Governance must be light enough to allow speed, but it must be completely visible.
00:18:44.319 --> 00:18:52.400
The AI has to document every single inferential step, cite every external data source, and preserve its rationale in plain text.
00:18:52.640 --> 00:19:00.319
So when the regulator inevitably knocks on the door and asks why a transaction was blocked or approved, you can show them exactly how the decision was made.
00:19:00.480 --> 00:19:03.599
You cannot have a black box in global compliance.
00:19:04.079 --> 00:19:08.799
Which brings us all the way back to that core theme of the summer compression: trust.
00:19:09.039 --> 00:19:16.799
You have to prove institutional readiness, which leads us to the final and honestly the most ironic realization of this entire deep dive.
00:19:17.119 --> 00:19:19.200
The human element required to sustain all of this.
00:19:19.519 --> 00:19:19.759
Yes.
00:19:20.000 --> 00:19:20.720
Think about it.
00:19:20.880 --> 00:19:31.279
After spending this entire conversation discussing hyperspeed global payments across borders, incredibly complex AI infrastructure, and algorithmic reasoning to catch international criminals.
00:19:31.440 --> 00:19:36.720
How do the executives actually building these massive futuristic systems do business with each other?
00:19:36.880 --> 00:19:39.759
According to the magazine, they are reverting to the 1680s.
00:19:40.000 --> 00:19:42.799
It is such a brilliant observation by the editors of the magazine.
00:19:43.119 --> 00:19:45.839
The industry is facing massive digital fatigue.
00:19:46.000 --> 00:19:55.039
The loud, chaotic trade show floors in Vegas or London, where every single booth slaps the word AI on their lanyard just to get a sliver of attention.
00:19:55.200 --> 00:19:56.640
They are losing their effectiveness.
00:19:56.960 --> 00:19:58.240
Nobody wants to go to those anymore.
00:19:58.480 --> 00:19:58.880
Exactly.
00:19:59.039 --> 00:20:01.839
It's too much noise, and everyone's guard is up.
00:20:02.079 --> 00:20:05.920
The new meta for making multi-million dollar deals is the table of ten.
00:20:06.240 --> 00:20:09.519
It's these small, highly curated private dinners.
00:20:09.680 --> 00:20:12.720
The stated topic of the dinner isn't even a sales pitch.
00:20:12.880 --> 00:20:19.359
The topic is just a good excuse to get the right ten people in a small, quiet room to break bread and actually talk to each other.
00:20:19.680 --> 00:20:23.359
The editorial draws a direct comparison to the origins of Lloyds of London.
00:20:23.680 --> 00:20:24.079
Yes.
00:20:24.400 --> 00:20:29.440
The global insurance market, Lloyd's of London, didn't start in a sterile corporate boardroom.
00:20:29.599 --> 00:20:36.640
It started in a coffee house in the 1680s because a man named Edward Lloyd curated the most accurate, reliable shipping news.
00:20:36.799 --> 00:20:46.480
By providing that niche value, he brought the smartest, most influential sea captains and merchants through the door, and the conversations just naturally turn into foundational business deals.
00:20:46.799 --> 00:21:05.519
If we synthesize everything we've explored today, whether we are analyzing the strict, visible governance of AI agents, Bunks' obsessive, borderless banking strategy for nomads, or even just how enterprise software is sold in 2026, the foundational currency of this entire era is trust.
00:21:05.839 --> 00:21:06.240
Absolutely.
00:21:06.559 --> 00:21:14.400
And trust simply cannot be automated on a crowded convention floor, and it certainly cannot be fast-tracked by burning venture capital on flashy ads.
00:21:14.559 --> 00:21:19.440
It requires operational discipline, absolute transparency, and genuine human connection.
00:21:19.839 --> 00:21:21.279
You can't fake maturity.
00:21:21.440 --> 00:21:25.519
The teenager has to actually grow up, put in the work, and build real relationships.
00:21:25.839 --> 00:21:26.240
Exactly.
00:21:26.480 --> 00:21:30.000
Well, thank you for joining us on this deep dive into the summer compression.
00:21:30.160 --> 00:21:35.119
It's a truly fascinating time to be watching the financial world quietly rewire its foundations.
00:21:35.279 --> 00:21:40.400
But I want to leave you with a single slightly haunting headline that was buried at the very end of our source material.
00:21:40.559 --> 00:21:41.519
It's simply read.
00:21:41.759 --> 00:21:44.480
When the buyer is an algorithm, who are you talking to?
00:21:44.799 --> 00:21:45.839
That's a chilling thought.
00:21:46.160 --> 00:21:46.559
It really is.
00:21:46.720 --> 00:21:56.880
As agentic AI gets smarter and these systems start taking over corporate procurement, vendor analysis, and risk checking, what happens when human coffee houses aren't enough anymore?
00:21:57.039 --> 00:22:05.359
Will we eventually reach a point where we have to figure out how to make one AI system take another AI system out to a a nice digital dinner just to build trust?
00:22:05.519 --> 00:22:06.640
Something for you to ponder.
00:22:06.799 --> 00:22:08.079
We'll see you on the next deep dive.