PFM Problems, Super Apps, & Narrow Banking
Happy Wednesday, Fintech Listeners!
I hope your week is proceeding as expected. And if not, I hope it has been better than you thought it would be!
A quick PSA — I no longer like the NBA. It is stupid and I will not be thinking about it for the foreseeable future.
Thank goodness I have plenty of stuff in fintech to pay attention to!
Speaking of which …
— Alex
P.S. Most apps don't advertise their payment infrastructure, but embedded payments have become the backbone of modern financial products. And the gap between the companies who built it right and those who bolted it on is starting to show.
On May 27, I'm sitting down with Jay Dearborn, COO International at WEX, to talk about what "built it right" means: developer-first APIs, compliance infrastructure that scales, and a payments system that never goes down.
3 BIG IDEAS FROM THE PODCAST
This week on Fintech Recap, Jason Mikula joined me from Oaxaca, where he was technically on sabbatical and technically still working. We agreed it was the sabbatical fintech deserved.
We covered 25 years of personal financial management apps and whether AI finally breaks the ceiling, why there are no cats super apps in America, the PACE Act, and what prediction markets like Kalshi and Polymarket are truly building when you look past the tweets.
Tune in for the full conversation here
And read below for my three big ideas...
#1: Can AI Solve the 15% Problem?Copy anchor linkCopied
Personal financial management (PFM) has been working on the same Rubik’s Cube for 25 years. Mint tried. Monarch, Copilot, and YNAB are still trying (with a better business model, but the same UX as their predecessors).
PFMs, as a product category, have a ceiling problem. Somewhere between, say, 10-15% of people (or less!) proactively want to categorize their transactions, build budgets, and review their financial dashboards on a monthly basis. All other consumers say they want help with their finances and then, in practice, don’t want to do any of the work.
Jason made this vivid with the example of watching his father record a transaction in the check register of a physical checkbook. He wasn’t ever a Mint user, but, spiritually-speaking, he kinda was. His brain works that way. He’s part of the 15%. So is Jason.
This is why the first generation of PFM failed. Mint attracted the people who already wanted this. It didn’t reach the 85% who aspired to manage their money better but had no real intention of sitting down to do it.
Now there's a third generation forming: Perplexity with an integration with Plaid, OpenAI's acquisition of Hiro (the company that Ethan Bloch built after Digit), and the embedded AI agents from Cash App, Revolut, and Public (etc!). The premise is that natural language interfaces might be a more accessible entry point. If you can ask a chatbot a simple question about a charge rather than opening a dashboard, maybe the appeal widens. I think that's probably right on the informational side.
The more difficult question is whether any of this can change behavior rather than just report on it. That’s where every previous generation of PFM fell short, and it's where the real value lives.
Perhaps the future of PFM is agentic?
🎬 DIRECTOR'S COMMENTARY
Credit Karma is an interesting example in this discussion, and one that Jason and I spent a few minutes talking about.
In some senses, it was a part of the PFM 1.0 generation. It had the same business model as Mint, but functionally it was aimed at a slightly different target (credit score management and loan shopping) which, I think, led to more success and, ultimately, a better exit.
#2: The Super App FallacyCopy anchor linkCopied
Bolt raised $355 million at an $11 billion valuation in January 2022 to own one-click checkout (it feels absolutely crazy to type those words). By early 2025, Ryan Breslow had returned as CEO after a dramatic few years, announcing a pivot to what he called a financial super app.
Jason did the reporting, and what he found was not encouraging. The Google Play store showed roughly 5,000 downloads. By Jason’s account, the onboarding flow was riddled with obvious UX errors and was borderline unusable. And while Bolt was terminating most of its contractors and laying off about a third of its staff, Breslow was on a stage in Las Vegas discussing whether consumers want a single platform for shopping, payments, identity, and rewards.
Consumers do not. And the reason they don't has nothing to do with Bolt specifically (though Bolt is a mess and I doubt it could have succeeded in any universe).
American consumers have been clear about their preferences. They will use 10 different apps if each one is better than anything else for that specific job-to-be-done. The idea that you can build a bundle from scratch and convince people to consolidate onto it requires that every product in the bundle be best-in-class at launch, and no company has ever done that.
The most consistently successful way to build in fintech is to identify one wedge, one thing that banks should do better, do that, and then move onto the next problem. SoFi started with student loan refinancing and slowly built outward. Robinhood started with commission-free stock trading. These aren’t super apps in the sense the term usually gets used. They're sequentially rebundled consumer banking providers, with each new part of the bundle earned rather than declared.
When a founder announces a super app, what I hear is: I want what JPMorgan Chase has, without any understanding of how JPMorgan Chase got there.
Banks are, functionally, super apps. But they got there over decades of building trust in one product at a time (and using that trust to earn access to the next one).
Shortcuts don’t work. Bolt is proof of that.
#3: Narrow Banking … In Many FormsCopy anchor linkCopied
The PACE Act probably won't pass (Congress being Congress and all). But its underlying logic isn't going away.
The bill would create a new category of registered provider: non-bank companies that hold money transmitter licenses in at least 40 states, maintain one-to-one reserves, and meet OCC risk management requirements … and give them a path to direct access to the Federal Reserve’s payment infrastructure.
The 1:1 reserve requirement is interesting: it means you can’t lend out the money you're holding. You can move money. You can hold money. But you can’t lend money (or do anything else risky with the customer money you are holding onto).
That is, functionally, narrow banking. And it's arriving from multiple directions simultaneously.
Stablecoin issuers under the GENIUS Act are being asked to operate roughly the same way. The national trust bank charter applications flowing into the OCC are also pointing in a similar direction. What you see, across all of these different legislative and regulatory threads, is steady movement toward carving payment and custody functions away from credit creation.
The banking lobby is working very hard to prevent it, primarily on the stablecoin yield question (banks are, apparently, not thrilled with the latest compromise, but they may be running out of room to negotiate something better).
The banks want to fight that specific battle. But they might be missing the larger one. Narrow banking isn't just happening in crypto. It's happening through fintech companies, through the PACE Act, through bank charter applications from nontraditional players … you get the idea.
WHAT I'M LISTENING TO
#1: Why Too Much Freedom Is the Enemy of Success (Plain English) 🎧Copy anchor linkCopied
I could not be more excited to read David Epstein’s newest book, but if you’re not familiar with it or David’s prior books (which are great!) I’d recommend listening to this conversation as a primer.
#2: Reintroducing The Fintech OGs (This Week in Fintech) 🎧Copy anchor linkCopied
Two really interesting guests in this podcast episode. Highly recommend adding this one to your queue!
Thanks for the read! Let me know what you thought by replying back to this email.
— Alex
