đ Welcome back to another edition of Fintech Radar!
If youâre new, here is a breakdown of what you can expect from each issue.
If you missed our recent editions, you can catch up here. Some previous issues you might want to check out if youâre new include âA Deep Dive Into The Cash Appâs Growth Machineâ, âThe Future Of Payment Initiationâ, and âCurrent: Doing It Differentlyâ.
Your ad could be HERE!
Fintech Radar is a must-read for fintech founders, operators, and investors. If thatâs your target audience, placing an ad right HERE is a cost-effective way to reach them.
If this sounds like a good fit for your brand, head to our sponsorship page for more details and to secure your advertising slot. Prices start at $100 per issue!
If you have any questions, reply to this email and ask away!
đ€ Cloudflare gives AI agents an identity and a wallet, Fortune
đ The Rundown: Cloudflare announced Cloudflare Wallets and cloudflare.pay last week during its Agents Week. Account holders get an Account Wallet holding stablecoins, and can issue Virtual Wallets to individual agents that spend through API keys within limits the owner sets. Payments run over x402, the open protocol Coinbase originated. Currently handle reservations are live, and the product is the buy-side counterpart to the Monetization Gateway Cloudflare announced in July, which lets sites charge agents per request.
đ„Ą Takeaway: Everything is fintech... eventually. Every technology company is ultimately tempted into making a payments play, and it all too often follows a similar pattern. Money flows in around the business as it continues to grow and one day someone works out that none of that is being monetised, and poof, a payments product suddenly appears. Most of the time the fit is poor and the thing reads as a line item somebody needed for a board deck rather than a product the company had any right to build.
This one might be the rare case where timing and the market had more to say than a simple âletâs get some of that actionâ. In Cloudflareâs case, it sits in front of a very large share of the webâs traffic (like a very large share), and what it has always sold is deciding what happens to a request before it reaches a server. Charging for a request, and funding the software making the request, are the same shape of problem as the one it already solves. Nobody had to invent a reason for Cloudflare to be here, but the timing, with agent payment standards starting to take shape, was likely a decisive factor for them to act.
By Agent Economyâs count, x402 has settled roughly 160 million payments worth about $41 million to date, which works out at something like 26 cents each. An enormous number of tiny payments for API calls, inference and data, and nothing that would trouble a mid-sized processor. Letâs be clear: itâs still very early. Historically, micropayments have failed on the consumer web for thirty years because a human will not stop to make a decision worth 26 cents. Agents have no such problem.
Serving both sides is the interesting part. Monetization Gateway prices the request, Wallets funds the buyer, and Cloudflareâs network is the thing deciding whether the bot request gets through at all. Two-sided markets tend to die in the gap where neither side turns up first, and building both ends yourself is the standard way through it, because you can seed one side until the other bothers to show up. Iâd also guess they are not yet certain which side is the valuable one, and owning both is a reasonable way to put off having to decide, and a handy bootstrapping mechanism for the whole segment.
My guess is this settles into two separate markets rather than one. The card networks and Stripe are building agentic checkout for the case where an agent buys something a human would have bought, with a card and a chargeback sitting behind it. x402 is for the case where software pays software for a resource nobody would ever have opened a browser for. The consumer end is where all the action is at the moment (e.g. agents acting for humans buying tickets to concerts etc.). The machine-to-machine end is where I would expect the transaction count to end up, and Cloudflare has quietly decided to build for that future.
đŠ Plaid and Sierra put a bank account inside the agent, Sierra
đ The Rundown: Plaid and Sierra announced a partnership last week that lets AI agents built on Sierra ask a customer for permission to connect their bank account inside a conversation, then use that data to carry the job forward without the customer leaving the chat. Sierra launched its Horizon agents in July, built to pursue long-running goals such as originating a loan or settling an insurance claim across multiple sessions. Sierra says the agents run compliance checks with defined points for a human to step in.
đ„Ą Takeaway: It seems Plaid is going deep on agents. I covered the first half of this back in issue #160, when OpenAI put a PFM experience inside ChatGPT and every one of those account connections ran through Plaid, across more than 12,000 institutions. That one was read-only. Connect your bank, your cards and your brokerage, then ask the model questions about your own money, which is roughly what a standalone PFM app used to do with a great deal more engineering. Sierra is the same piping but with a more B2B use case in mind.
Where the ChatGPT version answered questions, Sierraâs agents are trying to get a job done. A borrower who does not clear on credit score alone gets their cash flow reviewed inside the application rather than waiting for someone to pull statements. A claim that stalls on a bank detail keeps moving. The customer is also not sent off to a portal and told to come back, because Plaid Link surfaces inside the conversation itself, and anyone who has watched drop-off rates on a redirect knows what that is worth.
Plaid spent a decade as the connection between a budgeting app or a lender and somebodyâs bank account, and the value sat in bank coverage nobody else wanted to go and build. That coverage has not changed. What changed is who needs it. Every agent that wants to do anything useful with a personâs money needs the same plumbing, and Plaid arrives with ten years of experience deep in the pipes. Bloomberg reported in July that it has been holding early talks about a US IPO, and whenever that lands, the story it tells will be about agents rather than just open banking. How times have changed.
đŒ Monzo tops one million business accounts, Finextra
đ The Rundown: Monzo passed one million business customers last week, which it says makes it the first British digital bank to reach that number. Monzo Business launched in 2020, and the customer count has doubled in the past two years. One in six UK businesses now banks with Monzo, rising to one in five newly formed businesses. More firms used the Current Account Switch Service to move to Monzo during 2025 than to any other business bank.
đ„Ą Takeaway: A business current account is a sticky product. It is wired into the accounting software, the payroll run, the card terminal and a decade of direct debits, and the pain of moving it usually swamps whatever you save. Barclays, HSBC, NatWest and Lloyds have leaned on that inertia for a very long time. But when a founder decides enough is enough and/or the economics bend, businesses move.
Business banking behaves very differently from the consumer side in that respect. Consumer banking runs on brand, habit and whoever your parents banked with. A business will move if you show it a better outcome, because the person deciding is the same person doing the bookkeeping at nine on a Sunday night. Monzo has spent the past year putting a tax pot, a built-in tax filing tool and sole trader pensions with Inclusive Money into that account. Very little of it is banking. It is the admin around running a small business, which is the job the customer is trying to finish.
One in five newly formed businesses is the figure I would be staring at if I ran the SME business at one of the Big Four in the UK. Companies that start with you and grow tend to stay, so losing the formation moment is expensive in a way that takes years to show up. It has not hurt the incumbents yet, because a two-year-old company with ÂŁ4,000 in the account is not where their revenue comes from, and Monzo pointing the whole company at this after walking away from the US earlier in the year suggests it knows exactly what it wants to rally around.
Having said all this, itâs worth noting a million business accounts is not necessarily a million valuable ones. A large share will be sole traders and micro-businesses with thin balances and no borrowing, while the incumbent franchise sits in lending to established SMEs and the deposits that come with it. Monzo has the licence and the deposit base to move up into that, and I could see it happening, though taking real SME credit risk in a soft UK economy is a very different business from selling tax pots.
đ Nu MĂ©xico switches on as a bank, The Rio Times
đ The Rundown: Nu MĂ©xico began operating as a full bank last week, completing its conversion from a Sofipo to a banco mĂșltiple after a technical migration the evening before. Mexicoâs CNBV granted the final operating authorisation in July, roughly three years after Nubank applied. More than 15 million customers moved across, and deposit protection rose from around 225,000 pesos under the Sofipo scheme to about 3.5 million pesos under IPAB. Nu says it is now the largest digital bank in the country and will stay branchless. Payroll accounts and a wider credit offering are next.
đ„Ą Takeaway: In Mexico the nĂłmina account is how a bank gets anchored to a customer (as is the case in all markets), and the incumbents have owned that channel for decades through employer relationships nobody has managed to prise open. A Sofipo could not offer one. Nu can now, and it arrives at the starting line with 15 million people already using it for something else.
The deposit insurance change is also good news. Going from roughly 225,000 pesos of cover to about 3.5 million changes what a person is willing to leave sitting in the account. Thatâs the line between where you keep your spending money and where you keep your actual savings, and Nu has spent seven years on the wrong side of it.
Three years from application to switch-on, and every market has its own version of that queue. The length of it is precisely why nobody has walked in and taken payroll off BBVA or Banorte before now. Nubank has said it thinks the Mexican business can eventually reach the size of the Brazilian one and that the focus is on becoming the primary banking relationship for as many Mexican customers as possible. Primary relationship in Mexico means the salary lands with you.
The branchless part is the bit I am curious about. Every incumbent still runs branches partly because payroll has historically involved an employer, a corporate banker and a great deal of paper. Itâs a bet on where things are headed in the market, and time will tell if it works the way they think it will.
đž Western Union launches Stablecard with Rain, Western Union
đ The Rundown: Western Union launched Stablecard last week with the stablecoin infrastructure player Rain, pairing a digital wallet with a Visa secured credit card. Customers receive Western Union transfers directly as USDPT, the companyâs dollar stablecoin, and spend the balance anywhere Visa is accepted. USDPT is issued by Anchorage Digital Bank on Solana and redeems 1:1 for dollars. Stablecard is live in 37 markets including Argentina, Colombia, Mexico and the Philippines, with more than 60 targeted by the end of the year.
đ„Ą Takeaway: A fortnight ago Western Union was winding down the European digital bank it launched in 2022. Now it is handing out a dollar wallet and a Visa card across 37 emerging markets. Those two things look like more or less the same product, but when you pull back the strategy and execution curtains theyâre not the same business at all.
The European version made Western Union the bank. The money sat on its own balance sheet under its own Austrian licence, with deposit protection, capital requirements and a supervisor attached. The job that account was doing was turning a remittance customer into a current account holder: salary in, bills out, a card, somewhere to save. A deep relationship, slow to build and expensive to run, and the payoff only arrives once the customer treats you as their main bank.
Stablecard does none of that. Western Union is not the issuer, since USDPT comes from Anchorage Digital Bank, and it is not the card infrastructure either, since Rain supplies the wallet. The value sits in a token somebody else mints, against reserves somebody else holds, spent on a card somebody else issues. Western Union brings the brand, the corridors and the customer. No balance sheet exposure, no deposit insurance, no net interest income. The money comes from interchange and the spread, which is roughly the business Western Union was already in.
The difference feels marginal from the outside and is anything but. The bank wanted to be the place your salary landed. Stablecard only wants to be where the money sits between arriving and being spent, which for someone receiving a remittance in Buenos Aires or Manila is the job that actually needs doing. Nobody there is going to pay for groceries in USDPT. They are going to tap a card that settles from a USDPT balance, at a merchant who never learns any of this happened, and Solana is plumbing, which is generally where a blockchain has to end up before it is any use to a normal person.
The Defiant counted about $7.4 million of USDPT in circulation across 162 addresses on the day Stablecard launched, which is a rounding error against the volume Western Union moves. The product is also described as a secured credit card without much explanation of the credit mechanics, the limits or the fees, and I would want to see the pricing before deciding the receiver comes out ahead. The direction is right regardless. A dozen fintechs across Latin America and Africa have already built dollar accounts for people whose currency is sliding, and Western Union is late to it, but it turns up with the agent network, the corridors, and the person on the receiving end already in its records. Looking at the pieces theyâve assembled, this feels right, but letâs see how they execute it as a âlegacyâ brand.
đ° Rivo raises $3.1M and comes out of beta, Rivo
đ The Rundown: Rivo came out of beta last week alongside a $2.7 million seed round, taking total funding to $3.1 million. Backers include South Park Commons, Wisdom Ventures, Script Capital, 645 Ventures and 20VC, with former Nubank chief product officer Jag Duggal joining as an angel and advisor. Founded in 2025 by Ambrish Tyagi, previously at Cruise, Rivo connects to a customerâs existing checking account, moves idle cash into short-term US Treasuries through its banking partner Jiko, and returns it before the bills fall due.
đ„Ą Takeaway: PFM apps have spent fifteen years telling people things they already knew. Rivo does the thing instead. Tyagiâs line is that recommendation engines tell you what to do and Rivo just does it, which is the right instinct for a category that has never managed to turn insight into action. He came to this from Cruise and calls it self-driving money, which is a line borrowed from the last job but also a fair description of the product. You set it up once and it moves your money around without asking again.
Rivoâs own site puts its yield at about 3.7% against an average checking account at 0.07%, and This Week in Fintech reports the fee at five basis points a month on average daily balance. Call it 60 basis points a year, so roughly a sixth of the yield it generates. A legible price for a real service, which is more than you can say for the interchange-and-hope model most consumer fintechs launch on. It is also quietly taking deposits out of banks while leaving the account open. The customer keeps their bank, their card and their direct debits, and the balance that made the relationship profitable goes to Treasuries.
Sweep accounts have done a version of this for decades and almost nobody uses one, which is the inertia Rivo has named and is charging to solve. Whether people are comfortable letting something move their money around for them is the question this category always runs into, and it is not really a yes or no. Some will set it up and never look again. Plenty will not, at any price, however good the maths. The question is what the size of that market is, and maybe more importantly whatâs going to be the catalyst for that number to rise (at venture scale).
đ° Moment raises a $22M Series A, FinTech Futures
đ The Rundown: Cape Townâs Moment announced a $22 million Series A last week led by AlphaCode Venture Partners, with returning investors General Catalyst and MultiChoice and a first cheque from Canal+. Set up in 2023 as a joint venture between MultiChoice, Rapyd and General Catalyst, Moment builds payment collection and revenue retention tools for enterprise merchants across Africa, and says it processes around 600,000 transactions a day. Total funding is now $55 million.
đ„Ą Takeaway: The product is failed payments. Abandoned carts, debit orders that bounce, subscriptions that lapse because the collection did not go through on the day. In markets where a meaningful share of recurring payments fail on the first attempt, the money is in the retry logic and the recovery flow rather than anything happening at the checkout.
MultiChoice built it because it lives with the problem. A pay-TV business running across a couple of dozen African markets is, operationally, a debit order collections business with content attached, and a few points of collection rate is most of the margin. Building the thing you need yourself and then selling it to everyone with the same problem is an old and reliable way to start a payments company.
đ§ Mastercardâs Marc Pettican on the road to a $17.4 trillion virtual card market, Tearsheet (August 5)
Pettican runs corporate solutions at Mastercard and gets into why virtual cards keep growing in B2B, how you balance network economics when the buyer wants the float and the supplier resents the fee, and where the multi-rail story goes from here. Well worth a listen.
đ§ A New Intelligence Layer for Community Lenders with Mike de Vere, CEO of Zest AI, Fintech One-on-One (August 6)
De Vere has spent two decades putting machine learning into credit underwriting, and this one is good on what actually changes inside a community lender once the old scorecard comes out. Add this one to your playlist.
â€ïž Show Some Love For FR
đ§ Feel free to reach out if you want to connect. Iâm @alantsen on Twitter, or you can D.M. me directly by clicking the button below âŽ
Ps. If you like what Iâm doing with FR, please share it on your social disinformation network of choice. Iâd also appreciate it if you forwarded this newsletter to a friend who might enjoy it.
đ What did you think of this weekâs issue of FR?
I Love It! â I Like It â Not Bad â I Donât Like It â Itâs Awful