Happy Monday, Fintech Takers!

Today’s newsletter comes from the mountains of Montana, where I am co-hosting (along with my friends at EJF Ventures and the inestimable Dan Murphy of Allon Advisors) the Agentic Readiness Summit, a small, highly curated two-day event where we will be discussing all of the ways in which agentic AI is going to break the financial services industry (and what we need to do about it).

I’m already learning A TON and while the event is being held under the Chatham House Rule, I will be able to give you a readout on some of the key ideas in Friday’s newsletter!

- Alex

Was this email forwarded to you?

Subscribe Here

Saint Jerome Reading by Georges de La Tour.

3 FINTECH NEWS STORIES

#1: Centralization for the Win

What happened?

zerohash launched a new product:

zerohash … today announced the launch of AUTH by zerohash, a paradigm-shifting onchain authentication and account-linking product. AUTH solves for customer usability of onchain movements including for crypto, stablecoin and tokenized securities. Customers do not have to copy alphanumeric addresses or make chain decisions.

AUTH offers 600+ connections across custodial and self-custodial accounts, representing approximately 250 million users across 190+ countries, including platforms such as Robinhood, Gemini, Coinbase (Exchange and Wallet), MetaMask, Phantom, and more.

To lead this next phase of growth, zerohash has also named Ryan Beaudry as General Manager of AUTH. Most recently, Ryan was Executive Vice President of Global Product & Operations for Open Finance at Mastercard.

So what?

Let's be honest with ourselves: Crypto is highly centralized. Most people hold their crypto at Coinbase, Robinhood, or some other custodian.¹ And yet we keep up the pretense of decentralization. Your Coinbase "account" still presents to the outside world as a wallet with a long string of alphanumeric characters that no human being could possibly remember.

It's a bit like having a Gmail account but needing to know the IP address of your friend's mail server to send them a message. Or imagine if every Venmo payment required you to type in the recipient's routing number and choose between ACH, wire, and RTP, with your money vanishing forever if you picked wrong.

Nobody would put up with that.

Crypto users put up with it every day.

So, if crypto is mostly a centralized business these days, why not go all the way and mimic the experience that consumers already know from TradFi? Link your accounts, verify they're yours, and move your money. That's the open banking playbook, and it's not a coincidence that zerohash hired someone from Mastercard's open finance business to run AUTH.

However, the part that I find most interesting is the long-term opportunity. If Zerohash can bind verified identity information to self-custody wallets, it will own a network asset that no one else has. Outside of travel rule requirements (more on that in a second), exchanges mostly infer who owns which wallets using blockchain analytics. That's probabilistic. AUTH makes it consented and verified.

The registry that zerohash can build with AUTH gets a lot more valuable if travel rule requirements around self-hosted wallets become more common or binding. They're already moving in that direction. Under the EU's Transfer of Funds Regulation, crypto-asset service providers must verify that their customer owns or controls a self-hosted wallet for transfers above €1,000. Today that typically means satoshi tests and signed messages.² Tomorrow, it could mean pinging zerohash.³

In that world, compliance friction becomes a competitive moat. Linked wallets move instantly. Unlinked wallets get holds, questionnaires, and manual review. Nobody has to mandate anything; the friction does the work.

Self-custody purists won't love this. A registry tying wallets to real-world identities is exactly what self-custody was supposed to avoid. But those folks were never going to link their wallets anyway. For everyone else, self-custody already ran through centralized on- and off-ramps. AUTH just makes that reality more legible and a lot more convenient.

#2: Shop Pay and Shopify Protect

What happened?

Shopify rolled all of its fraud and dispute tools into a single brand, Shopify Protect.

The new pieces include:

• Pre-authorization blocking. When Shopify is near-certain a payment is fraudulent, it declines it before the payment is ever sent to the issuer.

• AI-written dispute responses tailored to the issuing bank, the store, and the reason code, with carrier proof of delivery attached automatically.

• A Rule Builder app that lets merchants write custom risk rules and backtest them against the past 90 days of orders.

• Chargeback coverage beyond Shop Pay. Shopify already eats the loss on eligible Shop Pay orders disputed as fraud. However, it has said that it will be extending that coverage to orders not on Shop Pay.

So what?

Shop Pay is one of the most strategically important assets that Shopify owns⁴ and the company has, for years, reserved its best perks (like free fraud chargeback coverage) for Shop Pay orders.

That’s why it’s notable that Shopify is loosening its grip in two directions at once:

• Beyond Shopify. Shop Pay for non-Shopify merchants started back in 2021 with sellers on Facebook and Google, and has steadily widened since. This latest batch of product updates takes the biggest step yet: Shoppers can now use Shop Pay inside AI agents like Claude Duet, Grok Bot, and Instinct to buy from any brand, whether it runs on Shopify or not. In agentic commerce, Shopify doesn't need to own the storefront. It just needs Shop Pay to be the wallet the agent uses. Every one of those transactions pulls in more buyers and more data.

• Beyond Shop Pay. Chargeback coverage for Shopify merchants' non-Shop Pay orders makes Shopify more attractive to merchants who never cared much about Shop Pay.

The only way to read this is that Shopify is trying to grow its network as aggressively as it can. And the reason for its aggression is data (consumer identities, devices, cards, delivery outcomes, etc.). Shop Pay and Shopify Protect are both instruments for growing Shopify’s proprietary dataset, and fraud prevention is the clearest example of a product that gets better with every new participant. Shopify said as much in its own announcement: "Fraud is not a payments problem. It is a commerce problem." Shopify sees the store visit, the cart, the checkout, and the delivery, across millions of merchants and hundreds of millions of buyers. A card flagged on one store is already known at the next, which can let Shopify block it before authorization.

There are costs to this strategy. Covering fraud chargebacks for free on non-Shop Pay orders is real liability, and it only pencils out if Shopify is super confident in its pre-auth blocking.⁵ But it makes sense if you accept the premise (as Shopify seems to have) that the network is the lever, and Shop Pay and Shopify Protect are just the most efficient ways to sign people up for it.

#3: Real Estate Agents Are Becoming Gig Workers

What happened?

Homa raised a seed round:

[Homa] has raised $5 million in seed funding for a service that uses AI and licensed agents to help people buy homes, then returns part of the buyer’s agent commission at closing

Quiet Capital led the round, joined by Restive Ventures and Cambrian Ventures. Homa already operates in Florida, California and Texas, and the money will help fund its geographic expansion and the launch of a new mortgage business.

The startup began generating revenue in April, and revenue grew sixfold over its first six months, according to [Homa CEO Arman] Javaherian. Homes purchased through the company have ranged from $150,000 to $2.4 million.

The company has returned nearly $1.5 million to buyers to date, Javaherian said. The average credit is about $10,560, he added.

So what?

Interestingly, Homa didn't start out as a brokerage. The first version of the product used AI to help people buy homes without an agent at all. That version ran into a wall: Some listing agents (the real estate agents who help people sell their homes) refused to work with unrepresented buyers and some listing contracts let the seller's agent keep both sides of the commission if the buyer had no agent.

So, Homa got a brokerage license.

And in version two of the product, showing requests (when the buyer wants to tour the house) go out to a network of local agents, and whoever accepts first gets a flat fee for the tour (no commission if the buyer buys). Offers and negotiations go to senior agents (10+ years, 200+ closings) who are paid a flat fee per closed deal, regardless of price. Javaherian says these negotiators can each handle 250 transactions a year, versus about 5.5 for a typical agent.

There are a few things about Homa that I find fascinating:

• Homa is a good example of why the real estate market is so slow to change. There are A LOT of stakeholders and, often, it only takes one to stop some disruptive new idea from getting traction. It sounds like Homa learned this the hard way when listing agents refused to engage with unrepresented buyers. Those listing agents didn't need to ban AI-assisted homebuying; they just needed to not cooperate. Homa had to become the thing it set out to replace (a brokerage) in order to get paid.

• There is a lot of fat to be trimmed out of real estate transactions, and Homa’s model illustrates what that may eventually look like for real estate agents. Showings get dispatched to whoever's available and paid per trip. Negotiation becomes a specialist job paid on volume rather than on the size of the deal. For buyers, this is obviously better: You can see six homes on a Saturday with six different agents, and none of them get paid more for steering you toward a pricier house. For buyer's agents, though, it's a bit of a bummer. It’s very Uber-ish. Homa's model doesn't eliminate the buyer's agent, but it compresses the profession into gig work at the bottom and a small number of high-volume specialists at the top.

• Every dollar of Homa's credit comes out of seller-paid buyer-agent compensation. If sellers increasingly decline to pay for agents, or competition compresses their compensation further, or buyers start paying their own agents directly, the rebate pool will shrink.

• If Homa aggregates enough buyers as a broker, the logical next step is to help arrange mortgages. This is reportedly the plan (Homa plans to arrange loans through outside lenders) and it’s (roughly) the same bet that Rocket made when it bought Redfin for $1.75 billion last year: The brokerage is the funnel, the loan is the margin. This is a very logical direction for Homa to go (the buyer agent credit could be used to buy down the interest rate), but it brings some RESPA concerns with it. Currently, the zombified version of the CFPB doesn’t appear to be expending a lot of energy scrutinizing kickbacks and referral fee arrangements in the mortgage market, but that will not always be the case.⁶

Sponsored by MX

The Great Wealth Transfer has left the whitepapers for dinner tables and viral threads about inheritance anxiety.

By 2048, an estimated $124 trillion will change hands, in what will be history’s largest capital transfer.

For banks and credit unions, will your accounts stay?

If the next generation inherits a digital experience constrained by data locked in a decades old core...well, let's just say the guard changes whether or not your core does.

The good news is that MX builds on that core instead of replacing it, so modernization ships in months.

And they have a fantastic new blog piece on The Great Wealth Transfer, too.

Read it here

2 READING RECOMMENDATIONS

#1: The Last Signature (by Stuart Cook, Atomic Settlement) 📚

Great piece from Stuart.

Checks and purchase orders and letters of credit, oh, my!

#2: Agentic commerce is at an impasse (by Vignesh Sreedhar) 📚

If you want to know where we are, right now, on agentic commerce (and why everyone working on it feels a bit stuck), read this.

A 10/10 primer on the subject.

*Bonus: Small Business, Big Onboarding Problem (by me, with Taktile) 📚

Business onboarding is broken, and the cause is variety: small businesses differ too much for KYB to standardize. Our infrastructure needs to get better at accommodating their diversity, and I write about how specialized AI agents could be the unlock, one that radically changes the cost and friction of business onboarding for institutions and borrowers alike. Read it here!

*This rec is brought to you by one of our fantastic brand partners.

1 QUESTION FROM FINITY

There are a TON of interesting questions being asked in Finity (our digital community for fintech and banking nerds). I’ll share one question, sourced from the community, each week. However, if you’d like to join the conversation, please apply to join!

How would you rewrite the Electronic Funds Transfer Act (EFTA) to accommodate agentic commerce?

If you have any thoughts on this question, reply to this email or DM me in Finity!

Thanks for the read! Let me know what you thought by replying back to this email.

— Alex

Footnotes

1.Or maybe Binance … where I’m told that there might have been a few problems?

2.A Satoshi Test is literally the DeFi equivalent of micro-deposit verification in TradFi. But with a much cooler name.

3.The open banking parallels here are fascinating. Plaid built the foundation of its business on instant account verification, as a much faster and more convenient alternative to micro-deposit verification.

4.According to Shopify, Shop Pay has 400 million users, which is a number that absolutely shocks me. 400 million?!?

5.It's also bad news for third-party chargeback guarantee providers like Signifyd, Riskified, and NoFraud, whose core product Shopify just made a free feature.

6.For my RESPA nerds out there: Section 8 lets private plaintiffs sue for treble damages regardless, so legal liability for RESPA doesn’t entirely disappear when the CFPB pulls back on supervision and enforcement..