FintechTalks

The episode contains a handful of genuinely interesting ideas - notably that the revolution in finance will come from customers wielding AI rather than banks deploying it, and that cryptographic identity makes AI-to-AI fraud structurally harder than human-targeting deepfakes. However, the ideas are rarely developed beyond a sentence or two, and several minutes are lost to conference pleasantries and mutual praise.

“the revolution coming in financial services, but it's because of customers using AI, it's not because of banks using AI”

“AIs live in a world of keys and certificates and encryption, digital signatures. And like I can make a fake video of Jim Kramer saying, oh, you should buy Dave Birch's stock, but I can't make a fake digital signature of Jim Kramer. Like math doesn't work like that.”

The reframe of 'the customer's AI becomes your customer' and the argument that AI agents are actually harder to defraud via deepfakes (because they operate in cryptographic space) are genuinely counterintuitive contributions. The rest - DORA, PIX/UPI benchmarks, stablecoin rails - is fairly standard fintech conference discourse.

“this idea that you have the customer's AI becomes your customer, not the customer anymore”

“the primary users of stable coins are gonna be AIs”

Dave Birch is a well-regarded, long-tenured expert in digital identity and payments with real consulting work and published writing behind him, not a content-farm thought leader. However, by his own description he is an 'author, advisor, and commentator' - not an operator who has built or scaled a product - which limits the practitioner depth a B2B operator audience would most value.

“I'm an author, advisor, and commentator on digital financial services”

“I wrote a thing a few years ago for Wired magazine actually about AI and banking”

The episode names concrete anchors - DORA, PIX, UPI, Tempo, Scandinavia's MobilePay/Vipps, the 10x cost-per-nine resilience rule of thumb - but no hard data is actually cited: PIX adoption is gestured at ('to a first approximation, everyone in Brazil has PICs') without a number, and all claims rest on assertion rather than research or lived operator metrics.

“you've got things like DORA, uh the uh European Digital Operational Resilience Act”

“to a first approximation, everyone in Brazil has PICs, basically”

The host's questions are consistently broad and occasionally incoherent, never challenging a single claim or asking for evidence. Valuable airtime is spent on a conference-atmosphere check-in and an extended closing tribute, and the host's rambling AI setup question fails to land a crisp follow-up that could have deepened Birch's more original ideas.

“what what's what's what what's interesting to you at the world in the world right now in in our industry?”

“And and um shifting now to AI it is um yeah, you know, I I'm I remember as you know for when different technology platforms come out, you know, there there's a bit of the the platform itself”