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

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.