Hey Fintech Nerds 👋,

I finally did it. For a long time, I got the feedback “I love Brainfood, but it's too long.” So I’ve split it into two weekly newsletters. This edition. Sunday’s Things to Know. It has my takes on the news, 4 fintech companies, and some content you should check out. Then on Thursdays, it’s Rants.

Missed Thursday’s 📣 Rant? It’s here 👉 🧠 Why Zelle wants to launch a stablecoin 👀 

Forwarded this email? You can subscribe by hitting the button below. Want to get in touch? Hit reply to this email or the sign up email.

Here's this week's Brainfood in summary

👀 Things to Know:

💸 4 Companies:

Content Corner: Choosing to Stay Human

Wait, where is the 📣 Rant? It’s here 👉 🧠 Why Zelle wants to launch a stablecoin 👀 

1. Federal Regulators publish proposed CIP rules for GENIUS-compliant stablecoins

The Federal Regulators: The Financial Crimes Enforcement Network (FinCEN), the Office of the Comptroller of the Currency (OCC), the Federal Reserve, the Federal Deposit Insurance Corporation (FDIC), and the National Credit Union Administration (NCUA) proposed rules for GENIUS-compliant stablecoin issuers to maintain effective Customer Identification Programs (CIP).

Federal Reserve Vice Chair for Supervision Michael Barr is concerned “that the GENIUS Act regulatory framework does not do enough so far to address the risks of illicit finance conducted through secondary market transactions in payment stablecoins.”

The rules say stablecoin issuers (e.g., Circle) must maintain these CIP programs, similar to banks ( while on/off ramps are often MTL licensed entities under state supervision). This includes much of what you’d expect, like a dedicated compliance officer, written procedures, transaction monitoring, alerting, audits, and ongoing training.

Beyond what banks do, stablecoin issuers and ramps must be able to seize, freeze, and/or burn tokens implicated in sanctions and run a dedicated 24/7 sanctions program.

Because the proposed rule currently limits the CIP mandate to the primary on- and off-ramps, Barr sees a potential compliance vacuum. A bad actor doesn’t need to pass KYC with the issuer; they can simply buy fully compliant stablecoins on the open secondary market (like a decentralized exchange) from an unverified, pseudo-anonymous address, bypassing core identity checks.

Barr is also concerned that all freezing is ex-post (after the fact) instead of ex-ante (before). In theory, a stablecoin issuer could block anything throughout the decentralized markets.

🧠 There are real crime, sanctions, and geopolitical risks with stablecoins. Russia's ruble-pegged stablecoin, A7A5, was built with Promsvyazbank to evade Western sanctions. Elliptic reported it processed over $100 billion in onchain volume before being banned by the EU. The Russian exchange Garantex was using USDT for billions in illicit flows, and Cambodia's Huione Group was laundering scam proceeds.

🧠 What Barr is proposing is very different to how banks work with cash today. When you withdraw $100 from a bank, they don’t KYC that cash as it moves in circulation.

🧠 Barr’s requirement for faster freezing is supported by evidence. ZackXBT on Twitter highlighted $420 million in alleged compliance failures across 15 cases since 2022, in which the US-regulated stablecoin issuer was slow to act or failed to act on illicit funds. Including the Drift hack itself, where $232 million in stolen USDC transited Circle's own bridge for six hours with zero intervention.

(For the full background check out Poor AML will ruin stablecoins)

🧠 Which private company polices a decentralized network? Networks like Ethereum and Solana are validated by 20+ different companies and are globally distributed. This is a feature not a bug. It’s what allows stablecoins to get to last mile markets at low cost.

🧠 Still, I think we can do better with technology. At Tempo* we have block/allowlists that can be baked into the tokens as they’re minted. Meaning, a stablecoin issuer or bank could ensure that the token could never touch a blocklisted wallet ex-ante.

🧠 My fear is we start policing technology. The CLARITY Act, SEC, and CFTC have now largely established that code is not a regulated market actor. ERC20 (the standard most stablecoins use) doesn’t have block/allowlists (although some stablecoin issuers add them). But it easily could be a standard feature (see TIP20).

2. Coinbase registered an AI with the SEC as an investment adviser 

Coinbase launched Coinbase Advisor, an AI you talk to in plain English that reads your full Coinbase portfolio and account history, suggests actions across crypto, equities, and derivatives, and surfaces ideas you wouldn't have asked for. It's non-discretionary, so you still have to confirm every trade, and it's rolling out to Coinbase One subscribers in the US first.

🧠 The registration is the most interesting bit of this news. They set up Coinbase Advisors, LLC and registered it with the SEC as an RIA, plus the CFTC and NFA as a Commodity Trading Adviser, taking on fiduciary duty under the 1940 Act. Robo-advisers like Betterment have held RIA status for a decade; the new part is a conversational LLM carrying those credentials. Contrast this with Robinhood who’s AI is "educational, not advice."

🧠 Coinbase's own disclosure says the outputs "may be inaccurate or incomplete" and you are responsible the loss. Hence you having to confirm every trade I guess. But it’s still wild to me that an 85-year-old fiduciary statute now points at an LLM that can hallucinate. This hasn't been tested anywhere, and we don’t yet know what could go wrong.

🧠 Contrast this with the UK and EU approach. Coinbase read the US existing law, registered inside it, and shipped to customers. The UK is building a targeted support regime (PS25/22, live in April) and running supervised AI Live Testing cohorts for firms that want to trial exactly this. But trials are different to just shipping. That’s why the US often gets innovation first.

🧠 Today it only sees what sits inside Coinbase. Your bank, pension and other brokerages stay invisible, so it's a co-pilot for one account. A whole-life wealth manager needs every account, which is where open banking comes in. (Plaid? 👀)

🧠 The AI adviser is as much a subscription hook as a wealth tool. Advisor sits behind the Coinbase One paywall. Clearly there’s a push for recurring revenue after a tough Q1 with crypto down. Coinbase also wants to become the “everything exchange,” they already offer prediction markets, and just announced plans for tokenized stocks

3. Karta raises $15m for the dollar card for international travelers

Karta is a premium credit card for global travelers who want to pay in dollars as they travel. Local premium cards incur FX fees while traveling, but premium Visa cards have no FX fees for regular swipes. The high net worth, high asset population lacks access to meaningful rewards.

🧠 Stablecoins move in where the card issuers move out. Amex used to offer “dollar cards” but has been pulling back on that offering. Here you have a high-asset, high-income population that wants a good card, but can’t get it. Now they can.

🧠 The round was led by Galaxy Digital, but nowhere does the release say “stablecoins.” But they use Rain, the stablecoin issuer processor. That tells you where we are in the hype cycle. A “stablecoin rail” is just infrastructure now.

🧠 I wonder if Amex had too much risk. It’s quite hard to fully know your customer when they’re in another country and onboarding them for a large credit line. My former banker DNA worries about AML risk here.

🧠 Risk is a data problem. Your bank, like Chase, which also gives you a credit card, has way more data about you than a new card issuer in another country. They also have local credit reports, and consortia data they can use.

🧠 But you get zero innovation if you take zero risk. We need people pushing these products into market, and using new channels to service customers. For example, most of the servicing is via WhatsApp, and 24/7. Imagine there was an alert, or a dispute, the customer comms here is baked in.

Fintech Nerdcon 2026 is shaping up. In the background, I’ve been working on speaker gifts that I can’t wait for you to see. And possibly the coolest floor space ideas ever. There’s at least one London cab, and whiteboards, and… Well, you’ll have to wait and see. If you’re in agentic commerce, using AI in a regulated company, stablecoins, or prediction markets, this is where your peers are; this is where you learn. This is where every conversation is worth your time. 👇

1. Artificial Intelligence Underwriting Company - SOC2 for AI Agents.

The company has created AIUC-1 a certificate for enterprise AI agents. Once an agent has earned this certificate, enterprises can then buy insurance against agent-created losses or damages. The standard covers data & privacy, security, safety, reliability, accountability, and societal risks.

🧠 Once you have standards and insurance, enterprise adoption follows. Elevenlabs and Intercomm are clients. But the measure of this company will be can they get more AI companies to use the standard, and will more enterprises see that as a qualifier? 

2. Rem - Stablecoin tokenization infrastructure for institutions

Rem helps banks, asset managers, and licensed issuers build custom stablecoin implementations, or tokenize credit, treasuries, funds or other real world assets. Today it works on Tempo*, Solana, Base and Ethereum. They’ve supported inter-company reconciliation with stablecoins and tokenizing a treasury bill fund on a Tempo private zone. 

🧠 There’s no lack of companies that can help you tokenize (Paxos, Anchorage, Securitize and countless others). What makes Rem stand out to me is their use cases are very capital markets focussed, and their network selection is the ones you’d pick between if you were considering at scale capital market use cases.

3. Saris - UiPath for AI Agents for Banks and CU’s

Saris helps banks and credit unions automate up to 70% of back office tasks like document intake, compliance checks and audit trails. They provide a white glove service that securely integrates with existing core banking and back office systems, and claim a “9.8x cost reduction per workflow”

🧠 Small banks need AI too, and they need the white glove service to make it happen. The board Saris has assembled tells you a lot about how they plan to go to market. Leaders of smaller banks and credit unions. 

4. Rep AI - The sales rep chatbot for your e-commerce store

Rep AI reads 1000s of user behavioral signals to understand a users intent, and position the right question or support for a user who might be hesitant or unable to find what they want. The goal is to mimic the brick and motar sales person, to narrow product choices, or deal with FAQ style question without creating the click away from the SKU. It works across the website, email, social, and even in whatsapp.

🧠 AI is coming for e-commerce conversion. Conversion is the north star metric for any commerce. Rep says they deliver 10 to 30% uplifts, with a 50 to 70% reduction in support tickets and a 16% higher average order value (AOV). For some folks this will be helpful, for others it will be clippy. But right now, it looks like it works and that’s the key.

1. Choosing to Stay Human

The sheer volume of AI writing is leading to a certain type of rage. Ethan diagnoses that badly prompted writing produces very little meaning per word, and it often hides the real benefit of writing to the writer: thinking. Various education studies showed kids using ChatGPT thought they were learning more, but when tested, were no further ahead than peers. Cognitive surrender is when you let AI do the thinking, choosing to stay human is choosing to do the thinking. 

🧠 As someone who regularly experiments with AI, and has built workflows for social posts. I certainly see the pushback. The algorithms on LinkedIn are now ranking much more for depth of insight. Which is nice. I can do that. I welcome the end of slop. 

🧠 I’d still rather someone actually engage with AI, than act all snooty and look down on anyone who tries. The worst thing you can do with AI is not experiment with it. 

🧠 Also, be willing to be wrong, have egg on your face, and don’t judge someone for trying with AI. I’ve absolutely hit send on things that I didn’t love or needed more edits, or had too many AI tells in there. But I’ve learned a lot in the process.

Remember, if you're enjoying this content, please do tell all your fintech friends to check it out and hit the subscribe button :)

Want more? I also run the Tokenized podcast and newsletter.

(1) All content and views expressed here are the authors' personal opinions and do not reflect the views of any of their employers or employees.

(2) All companies or assets mentioned by the author in which the author has a personal and/or financial interest are denoted with a *. None of the above constitutes investment advice, and you should seek independent advice before making any investment decisions.

(3) Any companies mentioned are top of mind and used for illustrative purposes only.

(4) A team of researchers has not rigorously fact-checked this. Please don't take it as gospel—strong opinions weakly held

(5) Citations may be missing, and I’ve done my best to cite, but I will always aim to update and correct the live version where possible. If I cited you and got the referencing wrong, please reach out