Episode summary
Anatomy of an Agentic Transaction
Fintech Takes
· · 1 hr 15 min · Hosts: Alex Johnson · Guests: Dan Murphy
Summary
The episode uses agentic AI as a lens for rethinking the plumbing of financial services, especially as AI assistants begin to act, transact, and sometimes break rules on a user’s behalf. The host and Dan Murphy frame the debate as a sequence of questions: who is the agent, what is it allowed to do, and what happens when it makes a mistake? They argue that agentic finance looks less like a brand-new category than a new pressure test for old open-banking and payments problems.
A major theme is authentication and authorization. Dan Murphy says KYA is harder than KYC because an agent is a new abstraction layered on top of a person, and that challenge gets worse because the US lacks a unified digital identity system. He suggests the field will probably need a mix of consumer-controlled permissions, platform controls, and standards, while also pointing to the UK, Australia, and Brazil as places where regulators have already built standing-authority frameworks for payments.
On liability, both are blunt that consumer trust has to come first. Dan Murphy argues the instinct to shift losses onto consumers is usually wrong, both for protection and for business growth. The host adds that the industry is likely to spend years arguing over responsibility only after adoption is established, because without trust there may be no meaningful market to fight over at all.
Key takeaways
- Dan Murphy says KYA is harder than KYC or KYB because agents are easier to spin up, harder to pin down, and can disappear quickly.
- The host argues that consumer trust has to precede liability fights, because agentic finance will not scale if people do not feel protected.
- Dan Murphy points to the UK, Australia, and Brazil as examples of standing-authorization frameworks that could inform agentic payments.
- The episode suggests authorization will likely be split among consumers, platforms, standards bodies, and regulators rather than settled by one simple rule.
- Both speakers expect the toughest open question to be the gap between what the user intends, what the agent does, and what a provider believes was authorized.
Notable quotes
“we're back to the future, right? And so we've been here before with respect to the screen scraping fight”
“If you are delegating to let's pick on Muse to go out into the world and then transact or interact with financial services on your behalf, that is a little bit different”
“we suck at KYB. Like we're very bad at it”
“who gets to decide the scope of authorization for agents”
“the first the first purpose of this is to make sure the consumer gets made whole if something goes wrong”
How the episode opens
This episode is brought to you by Increase. Increase offers API first banking for ambitious technology companies. Coast, Gusto, and Ramp rely on Increase to process more than $500 billion annually.
If you're building the next great financial product and need a bank that gets it, visit increase.com. Hello. And welcome to the Fintech Takes podcast.
Today's episode is a great one. I was delighted to be joined by my friend and frequent podcast guest Dan Murphy, partner at Alon Advisors, and formerly of the CFPB. Dan, as you likely know, is someone I lean on a lot when it comes to interesting sort of policy questions, particularly policy questions that intersect with really fast-moving emerging trends within the Fintech ecosystem.
And of course, the fastest-moving and most disruptive trend in the Fintech ecosystem right now is AI, and in particular, agentic AI. These robots that are out wandering the industry trying to help consumers, trying to help small business owners, and in the pursuit of trying to help their users breaking all kinds of terms and conditions, scraping websites, taking actions, in some cases making mistakes, and the question that Dan and I wanted to wrestle with is what does that mean for the financial services industry? What does that mean for regulation?
What does that mean for policy? Obviously, Dan has a very deep background in open banking. He was at the CFPB and helped lead the initiative under the last administration to write the rules around open banking and 1033 in the US.
I am a passionate amateur on the subject of all things open banking and 1033. And a lot of what's happening now in agentic AI rhymes with the fights that we've been having in open banking for a decade plus. So, in some ways it's creating these sort of flashbacks and these sort of moments of recognition for people who've been deep in open banking.
From the episode’s captions: roughly the first two minutes, lightly cleaned. Keep watching on YouTube →
Topics
- agentic finance
- open banking
- kyb verification
- agent authorization and liability
- consumer trust and fiduciary law
Companies and tickers mentioned
- Allon Advisors
Watch the episode
Watch “Anatomy of an Agentic Transaction” on YouTube
Summary, takeaways and quote selection by Octagon. Quotes are short excerpts; the full conversation is in the episode.
More from Fintech Takes
-
The Blue Wave Hits Financial Services -
AI Agents, SoFi, & the OCC’s “Irrefutable” Report -
Fintech Takes x C&R presents: The Customer Isn’t the Liability -
Stablecoins IRL (Tempo Edition) -
Fintech Takes x C&R presents: The Business of Lending, Seen From Canada -
Not Fintech Investment Advice: Symphony, Spoolis, Piston, MoneyStack
Fintech Takes, this episode and its artwork belong to Alex Johnson. Octagon is not affiliated with the show; this page is an independent summary.
