Welcome to Fintech Brainfood, the weekly deep dive into Fintech news, events, and analysis. You can subscribe by hitting the button below, and you can get in touch by hitting reply to the email (or subscribing then replying)
Weekly Rant 📣
🧠 Wallet Wars Pt 4: The Personal Finance Agent.
The next wave of consumer fintech isn’t an app. It’s a personal agent. And nobody’s built the product yet.
This is the fourth part in a series on the Wallet Wars.
Part 1: Wallets become identity containers.
Part 2: Wallets become the home of AI agents.
Part 3: The coming bot authentication crisis.
Today's piece: Who builds the guardian agent?
My OpenClaw helped me do my Taxes
This is the first tax season where AI actually helped me.
I needed to find all of my email receipts. Because my OpenClaw agent already has secure(ish) read-only access to my Gmails, it automatically stores them in a folder when they land.
So I had a package ready to send to my accountant. A CSV of descriptions and categories, already populated. It used my Google Maps history to produce a car mileage report. (I’m getting it the Xero API next).
Sure, I could have done this with an LLM. But I would have had to grant it access. Search for the pieces. Generate the prompt. Wait for the output. The magic of the agent is that it was just sitting there. It had already done the work in the background. I could even have it send it to the accountant without me but I wanted to review.
This is what “AI-first fintech” actually looks like. An agent that completes tasks by learning your workflow and building around it.
The experience is invisible.
And invisible is the next big fintech opportunity.
Who’s building the personal finance agent?
Seriously, where is the consumer finance agent?
One question I get asked most often by VCs is why there are almost no consumer AI-fintech companies.
There’s one very large VC who believes “Fintech is Dead,” so may not be actively looking, and, well, you do you, Marc.
But I can’t fathom the idea that this $38.6 trillion industry that throws off $1.2 trillion in profit doesn’t have another major wave of disruption coming for how consumers interact with their finances.
The issue isn’t that consumer fintech is dead.
The issue is that we’re looking for an app. When the future of consumer finance is an agent.
The old saying goes that what the nerds are doing on the weekend is what everyone will be doing in 10 years. Right now those nerds are setting up their OpenClaw (and probably trying out peptides). Those agents are automating all kinds of rote work. Checking calendars for schedule clashes, running automations to log daily food, building workflows that turn a voice note into a social post.
My thesis: Personal agents will play a major role in the financial experience for the next decade. What we’re missing is the operating system for them.
What I don’t know is who will build it.
You Can’t Scale a Tamagotchi
The reason OpenClaw is still very much an enthusiast product is that it’s hard mode.
You have to either buy a dedicated Mac Mini, and/or set up a Docker instance on a virtual private cloud, then link various LLMs, skills, and plugins to your personal accounts and API keys. As I wrote recently:
That’s the thing about claws. You don’t buy them as a product. You raise them like children… The claw economy is built on the idea that you install skills.
Raising an AI like a child is a terrible consumer product experience.
Mainstream users don’t want a Tamagotchi. You can’t scale a Tamagotchi.
Sure you can pay some guy $6,000 to do it for you, but this is a product that should just work. It doesn’t.
You need a packaged, secure product.
Which is why the Nvidia CEO announced their own package for OpenClaw, NemoClaw and said “Every company needs a claw strategy.” When the CEO of Nvidia validates your thesis, we’re past the “is this real?” phase. The packaging has started.
You Are the Primary Key
There isn’t a “Plaid for everything ever” and there’s no single big company that sees your whole existence (and that’s a good thing).
Sure,
Google sees everything you’ve OAuthed into, but doesn’t have folder access to your local device, or your bank account data.
Apple has linked a lot of accounts, but doesn’t know what’s happening inside your bank’s core ledger or your Gmail.
Banks have your salary and KYC but can’t see your Amazon order history or your calendar.
Big companies can never access everything you do. There are lots of little silos of data.
But you can see it all today. You are the primary key.
Take X and OpenAI. Elon and Sam have a major beef. X doesn’t let OpenAI scrape its data. But me, as a customer of both, can poll the API for both.
Your personal agent can be that bridge. It sits on your device, acts on your behalf, and connects the silos that big tech refuses (or isn’t legally allowed) to connect. You just have to become a world-class security engineer if you’re going to wire up your finances to it. NBD.
Which brings us to the hard part.
From Agents to Guardians
Finance is high stakes. The place where your salary goes, your entire net worth, your bills get paid. This cannot fail. The consequences of failure are enormous.
Meanwhile, Claude Code will YOLO delete a database.
Which leaves two fundamental questions:
How do we make sure these agents are robust and resilient? Tested, highly available, and if something goes wrong, you’re made whole. Some sort of liability framework or clawback mechanism.
How do we make sure the agents are aligned to you? Helping you achieve better outcomes, not dragging you into prediction-market, engagement-bait, memestock slop?
Alex Johnson calls the second thing “guardian agents.”
Love. That. Term.
If I may extend the idea: a guardian agent is aligned to positive user outcomes, and is protective of the financial and security integrity of the user.
Building an agent that can help optimize your yield, shop around for better savings, and pay down debt faster is doable as a set of features. Making that a consumer-grade product is harder. You need three things to settle in first.
Trust. Can we verify this agent hasn’t been compromised? Can we monitor its behavior continuously, not just at login?
Identity. Which agent is this, and how is it tied to its human? We need to KYA — Know Your Agent — in the same way we KYC humans today.
Authentication. How does an agent prove it’s authorized to act on your behalf, across legacy systems that were built to verify humans, not machines?
I’ve written about this extensively in the context of wallets and bot authentication. The standards are emerging: agent tokenization from Visa, verifiable agent identities from startups like Caten Labs, new authentication protocols from the card networks.
As I wrote in The Checkout Page is Dead:
If the agent buys before the page even loads, every single thing we built our business on has to move upstream. Fraud detection. Identity verification. Conversion optimization. All of it. Upstream.
The infrastructure is coming along. Who wraps it into a product consumers actually use?
That’s the $38.6 trillion question.
Who Gets to Own Personal Finance — And Agents?
I’ve spent the past year writing about the Wallet Wars — the battle over who becomes the consumer’s center of gravity for payments, identity, and authentication.
The wallet will become the consumers’ center of gravity, not the account, app, or checkout. Consumers have countless cards, apps, and accounts that will be bundled by a handful of market actors like Apple, PayPal, Cash App, and possibly even Paze.
A decade ago, card issuers talked about being “top of wallet.” Now, it’s about being the wallet itself. Whoever owns identity wins.
Because whoever owned the wallet had your payment methods and identity together. And they’d be ideally placed to be the safe, secure home for your agent (as discussed in part 2).
I then evaluated each player on payment methods, identity, and trust. Apple was ahead on all three.
But that was the old scorecard.
The guardian agent adds three new requirements that reshuffle the leaderboard entirely.
Cross-silo data access. Can you see across Gmail, banking, brokerage, Amazon, calendar, and local files without becoming a centralized surveillance company?
On-device compute and privacy. Can the agent run locally, process sensitive data without sending it to the cloud, and keep the user in control?
Continuous trust and alignment. Can you monitor agent behavior in real time, flag anomalies, and ensure the agent stays on the user’s side?
The Wallet Wars were about who owns identity. The guardian agent war is about who owns agency.
Different game. Different winners.
The Personal AI Wars - Who gets to be your guardian agent?
If we go back to the wallet wars, I had contenders like Apple, Google, Banks, and Fintech companies. But now, we have the AI Labs, Open Source and New Startups all duking it out for a little bit of an unknown future. Now the picture looks like this:
Agent Wars - Who gets to be your guardian agent?
How do you even begin to make sense of that? Well there’s some new requirements
Identity: Do they own a government legal entity credential.
Payments: Do they own a payment method with acceptance
Authentication: Can they be used for strong authentication to almost any service
Data access: Can they see across your data silos to your whole life
AI Capability: Do they have a model or AI capability they own
Trust: Would you trust them with that data?
I’ve ranked the contenders by their relative strengths in six areas. ✅ is a clear win, ⚠️is a partial capability and ❌they currently have nothing.
Who? | Identity | Pay Method | Auth Method | Data access? | AI ability? | Trust? | P(win) |
|---|---|---|---|---|---|---|---|
Apple | ✅ | ✅ | ✅ | ⚠️ | ❌ | ✅ | 25% |
⚠️ | ⚠️ | ⚠️ | ⚠️ | ✅ | ⚠️ | 10% | |
Banks | ⚠️ | ✅ | ✅ | ⚠️ | ❌ | ✅ | 5% |
Fintechs | ⚠️ | ✅ | ✅ | ⚠️ | ⚠️ | ✅ | 10% |
AI Labs | ❌ | ❌ | ⚠️ | ⚠️ | ✅✅✅ | ⚠️ | 15% |
Open Source | ⚠️ | ⚠️ | ✅ | ✅ | ✅ | ⚠️ | 15% |
New Startup | ❓ | ❓ | ❓ | ❓ | ❓ | ❓ | 20% |
But Simon, why did you give a warning sign to blah and why does open source only have one tick and...
Here's my thinking.
Apple (25%)
Has a habit of being late and winning anyway.
They have the real wallet — payments, identity, passkeys, hardware security, the secure enclave.
The problem is they have no LLM and don't seem to be in a rush to get one. Siri is a punchline. Apple Intelligence underwhelmed.
But picture an open-source agent packaged for iOS with a frontier lab partnership. Apple controls the hardware, the secure element, the distribution. They could own this the way they own the smartphone OS. There's no rush.
Google (10%)
Has the deepest cross-silo data access of any company — your email, calendar, docs, search history, photos, location — and one of the leading foundation models. On paper, the best guardian agent builder alive.
In practice, they cannot get their shit together on payments or identity. Their ad business model is a fundamental conflict with guardian alignment. Every financial recommendation risks looking like a sponsored result.
Android is the default outside the US, so maybe. But I'm not holding my breath.
Banks (5%)
Are the agent of record for your money today. Your salary lands there. They KYC you. They make you whole when fraud happens.
They will be infrastructure for guardian agents — holding money, providing identity — but they will not be the agent itself.
Come on. COBOL. They took 10 years to ship a passable mobile app. Nobody's hiring their bank to be their AI.
There's an opportunity here for someone to repackage OpenClaw as a white-label product and sell it to banks. The way some banks buy vendor solutions for their mobile app.
Fintech Wallets (10%)
PayPal, Venmo, Cash App — are the most obvious candidates doing the least about it. 400m+ users, payment rails, NFC access, generational loyalty.
They're not wallets in the way Apple Pay is. That's a hard line to overcome.
AI Labs (15%)
Are closest to shipping the consumer agent. ChatGPT has ~800 million monthly users. OpenAI hired Fidji Simo.
Claude is slowly turning Claude Code into a personal agent that can do everything OpenClaw can, but slicker.
But they have no financial infrastructure, no banking license, no KYC, no regulatory moat. If an AI lab partners with the right financial infrastructure, that combination could be devastating.
But they're locked in an all-out coding agent dogfight right now and personal finance is not a priority.
Open Source (15%)
Is the most philosophically correct answer. The OpenClaw movement is this. Stability AI founder Emad Mostaque is trying to build exactly this.
No conflicts of interest. You own your data. You run it on your own hardware. But Linux won the server, not the desktop.
How does it get distribution? Remember, Android is technically open source. The "Android moment" is worth watching for.
A Disruptive Startup (20%)
Has a super solid chance of being the eventual answer. We're just too early for it. OpenClaw just happened. Nvidia just dropped NemoClaw.
This week I spotted Era, a personal MCP that lets you connect you bank accounts and money history simply to an agent so it can help you manage finances.
The infrastructure is assembling for something new to emerge — an agent-native company that builds backward into financial infrastructure the way Stripe built backward into the financial system.
So that right there is why nobody has built the finance agent. Because it's f*cking difficult. And we didn't even begin to unpack the edge cases and what could go wrong if your personal agent yolo deleted your life savings.
So whoever builds this has a heck of a task.
There isn’t a clear winner yet. We’re still super early.
In the guardian agent war, there isn’t a winner yet.
Maybe it’s a partnership. An acquisition. Or a company we haven’t seen yet.
One that starts with the agent-first assumption and builds backward into financial infrastructure. The way Stripe started with developers and built backward into the financial system.
That agent. That wallet infrastructure doesn’t exist, and almost everyone is looking for it.
The Wallet Wars were about who gets to be the wallet.
The next war is about who gets to be your guardian.
ST.
4 Fintech Companies 💸
1. Natural - Payments accounts for agents.
Natural helps businesses, consumers and agents move money between each other. The service has a dashboard with an overall “wallet” and homescreen to manage agents for procurement, expenses and billing. It helps orchestrate the payments that agents initiate for these B2B workflows. It exposes tool calls like pay, collect, bill pay, and transfer
🧠Is this just an FBO account with workflows? Probably yes. But there’s immense power in those workflows being packaged as tool calls. Like everything in “agentic” the right to win comes from being able to serve the new customer. The agent. A client money or “for benefit of” account is the magic that runs much of Fintech. But it’s usually packaged as a Fintech app or Neobank. Speaking of, these guys went out and got every angel. The CEO of Ramp, Mercury, Vercel, Figure. Ridiculously stacked. And B2B agent flows are the right starting point. This is the PSP of tomorrow (similar to Cantena labs).
2. Pay AI - Biometric approval for agent payments
Get Pay avoids giving agents card credentials by sending a request to the via SMS or push notifcation, and when the user approves with their biometrics or the OTP, Pay AI funds a virtual card and uses that to pay.
🧠This is a neat package for virtual cards. Reminds me of AgentCard but with a lot more thought about the security limits at the API layer. The product is waitlist but there’s a free version for tinkerers. Interesting entry point. This pattern feels right and where virtual cards and agents will land.
3. Orca Fraud - Sardine for emerging markets.
Orca is a platform for fraud detection at customer onboarding and transaction time, it provides investigation tools and case management. It has a full rules engine and anomaly detection capability. The team is based in Cape Town SA, and is live in more than 70 markets.
🧠Mobile money has different patterns. Not all fraud works the same. Models trained on fraud patterns in G20 countries often fall down when used for mobile money. Orca just raised its seed round to fix this.
4. Axiom Trust - The tech-forward fiduciary trust company.
Axiom Trust provides professional trust administration for families and beneficiaries, working with attorneys, RIAs, CPAs, banks, and family offices. AI handles document parsing, data normalization, monitoring, and reporting so fiduciaries focus on judgment. It also produces reviewable decision packages for actions like distributions with clear approval trails.
🧠 How many trust companies are leaning into AI? The idea of taking a highly regulated, sleepy part of the industry and rebuilding it to be tech and AI first is great, but needs founder-market fit. Former Sydecar founders, lawyers and deep infrastructure experts executing in a pain point you probably missed.
Good Reads 📚
The prediction market pitch is genuinely compelling: crowds aggregate information, prices reflect reality, and markets are better arbiters of truth than any commentator or expert ever will be. I believe that. What I don’t believe is that the industry can keep sidestepping its most obvious flaw and expect it not to matter.
That flaw is insider trading.
🧠 Snappy, enjoyable read from Santi on prediction markets. Needless to say, I agree.
That's all, folks. 👋
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
1
