Undercut Falling a Fir Tree by J F Ford.


#1: Credit Cards Are a Poor Wedge Product

What happened?

Petal, the fintech credit card company, is being acquired:

Empower has entered into a definitive agreement to acquire Petal and expects the acquisition to close by the end of the second quarter

Its planned acquisition of Petal, a New York-headquartered consumer credit card and FinTech company, will expand Empower’s products to include the credit card category in the U.S., the release said.

It was reported in November that Petal was seeking a buyer amid doubts about its viability and had received multiple acquisition offers. Consumer lenders like Petal had been hurt by high interest rates driving up the cost of borrowing, the report said.

So what?     

The terms of the acquisition weren’t disclosed, but I’m guessing the number wasn’t enormous. According to a report, Petal had been searching for a buyer for a while, and the company had done a couple of rounds of layoffs over the past couple of years since raising $140 million in January 2022 (the company raised more than $300 million total).

This isn’t a commentary on Petal itself, which from everything I can tell, has built a reasonably good business (compelling product, fairly low loss rates, beautiful logo).

It’s more of a statement generally on credit cards as a B2C fintech wedge product.

You – a venture-backed startup without a bank charter or a cheap source of funding – either go after underserved consumer segments that traditional issuers have been ignoring as Petal did (these consumers tend to be highly risky and not big spenders) or go head-to-head with the big issuers for prime and super-prime consumers as X1 did (as we’re seeing with the resulting Robinhood Gold card, this is a very-well-optimized market that doesn’t leave a lot of headroom unless you’re willing to lose a lot of money on your card).

Neither path has worked out well in recent years (though it may end up working well for the acquiring companies… we’ll see).    

#2: How to Be a Banker

What happened?

A startup focused on bringing generative AI to banks raised a Seed round:

A startup called Hapax, armed with $2.6 million in seed investment, launched Wednesday with a generative AI tool aimed at the financial services industry. The platform is designed to leverage generative AI to deliver industry-specific information, decision-making capabilities, and usable assets to banks and similar organizations.

So what?     

Hapax is interesting.

From what I can tell, they have essentially built their own LLM foundation model, exclusively trained on 13 years of proprietary data sourced through a partnership with CBANC, a Texas-based network of more than 8,500 financial services institutions and 600 solution providers. The dataset includes more than 20,000 documents, 10,000 hours of videos, and 230,000 conversations between bankers that include questions and validated answers.

The product was specifically designed to help close what Hapax calls the “information-access gaps” between big banks and small banks: 

Small and midsize banks and credit unions, which can’t invest as readily in skilled employees and technology as their larger brethren, rely greatly on consultants and what Hapax calls a “best guess approach” to run their complex businesses and stay compliant.

It’s not clear to me why Hapax’s solution is going to be meaningfully better than the “best guess approach” offered by consultants (LLMs are rather fond of guessing, too), but apparently, this tool really is meant as a general information ‘oracle’ for community banks. You can ask it to do everything from drafting a compliance policy that will meet regulators’ expectations to conducting a strategic analysis of your customer acquisition strategy.

Join Fintech Takes, Your One-Stop-Shop for Navigating the Fintech Universe.

Over 36,000 professionals get free emails every Monday & Thursday with highly-informed, easy-to-read analysis & insights.

This field is for validation purposes and should be left unchanged.

No spam. Unsubscribe any time.

Two observations:

  1. This LLM is trained on data from documents and conversations shared by a bunch of community banks and other financial service providers in Texas. No offense to those specific companies or to the great state of Texas, but is that really a dataset that is going to reveal the secrets of how to be a great banker? Now, if Hapax had trained its model on JPMorgan Chase’s internal data, we’d be having a different conversation.
  2. Much like how teachers are concerned with their students outsourcing their research, writing, and critical thinking to ChatGPT, I’m not sure I’d love this if I were a bank examiner at the FDIC or OCC. I want you to show me that you understand how to run a bank, not just prompt an LLM

#3: Seriously?

What happened?

A crypto thing raised some money:

Mezo, a Bitcoin scaling network built by Matt Luongo’s Thesis, has emerged from stealth with $21 million in funding.

Thesis calls Mezo a Bitcoin “economic layer,” which will focus on creating an ecosystem of applications based on users’ economic needs, from groceries to tuition.

“Our goal with Mezo is to extend the Bitcoin network to bring 25% of the world’s economy on-chain — roughly in line with the size of the US economy today,” Luongo said.

So what?

Mezo wants to bring 25% of the world’s economic activity – $25 trillion – on-chain, through an “economic layer” that will feature an ecosystem of applications for helping consumers with stuff like groceries and tuition.

That’s … ambitious.

How exactly will this “economic layer” work?

Mezo uses “proof of HODL” as a consensus mechanism. Users secure the network by locking BTC and MEZO tokens. “The focus is aligning Mezo economically with BTC holders, who stake and earn rewards for running the network,” Luongo said.

Mezo has opened up bitcoin deposits and has already reached over $26 million in total value locked, according to its website. Mezo said the longer bitcoin deposits are locked by users, the greater their HODL score will be. HODL score is a points program.

Users will receive five one-time invitations to extend to their friends upon joining Mezo. Subsequently, Mezo will distribute reciprocal earnings based on the duration and amount of their friends’ bitcoin deposits made through the invitations.

So, you give them bitcoin (and buy their MEZO token), and you earn rewards, with extra rewards doled out if you recruit additional friends to buy into the scheme ecosystem.

And I guess the grocery and tuition applications just appear, at a certain point?

Stuff like this continues to make it difficult for me to take crypto seriously. 


#1: A data-driven look at fintech adoption in vertical software (By Matt Brown)

Matt writes some of the best content on the trend of vertical software + embedded finance, and now he’s publishing data on exactly how (and how much) vertical software companies are embracing this trend. I won’t spoil his conclusions here. Go read the piece (and subscribe to his newsletter if you haven’t already). 

#2: Chase’s New Advertising Offering Is A Stroke Of Genius (By Ron Shevlin, Forbes)

Ron’s analysis on this is great. Chase’s move has massive upside (his point about bundled pricing was particularly interesting). The challenge will be making its digital properties shopping destinations. Klarna has cracked the code on that. We’ll see if Chase can, too. 


Would you like to read a Fintech Takes research report that evaluates technology vendors in a specific area (e.g., application fraud detection, credit decisioning, collections)?

I feel like there’s room for something better (and less boring) than what Forrester and Gartner produce, but I would be curious to hear your thoughts. Would you read it? Would you pay for it? What technology areas would you like to see covered?

Alex Johnson
Alex Johnson
Join Fintech Takes, Your One-Stop-Shop for Navigating the Fintech Universe.

Over 36,000 professionals get free emails every Monday & Thursday with highly-informed, easy-to-read analysis & insights.

This field is for validation purposes and should be left unchanged.

No spam. Unsubscribe any time.