🧠 Will AI Kill Fintech SaaS?

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 📣

🧠 Is Fintech SaaS Dead?

Last week, I watched 10 agents build a production wallet app. In parallel. While I tweeted about Fintech news.

In the week Brex, Brex’ited for $5.15bn, Claude Co-work automated expense management as a feature.

a16z thinks AI kills SaaS, because there’s an explosion of app development. What YouTube was for video creation, AI is for software:

The stock market thinks AI kills SaaS.

DocuSign is down 40% from its peak. Workday cratered. The logic seems obvious: if anyone can vibe-code a competitor in a weekend, what's your moat?

Here's what I think they're missing.

I can spin up agents to rebuild Plaid's API wrapper in a day. I cannot spin up agents to renegotiate JP Morgan's data agreement. I can generate a fraud scoring model by Tuesday. I cannot generate 10 years of labeled fraud data to train it on.

Agents commoditize code. They don't commoditize network effects, data moats, or the 1000 idiosyncratic edge cases that only matter when you're operating at scale.

The last 15 years of Fintech infrastructure were built on companies like Plaid, Marqeta, and Sardine* that made building a bank-like app 10x faster. The next 15 years will still need them—but they'll have to ship 10x faster too.

The power of agents building products.

The velocity at which new, high-quality products can be built is staggering.

A powerful example of product velocity is Claude Co-work, which Anthropic built in two weeks. 100% with Claude Code. This is a multi-billion-dollar enterprise shipping a massive product in weeks. If that doesn’t send shivers down your spine, it should.

Don’t think of code generation.

Or text generation.

The power comes from having multiple agents work as a team. 

• A “manager” to orchestrate the team of agents

• Sub agents per logical division of labor (e.g. UX, platform, back end)

• Security audit agents

• Code refactoring agents (making the code performant)

Cursor built an entire web browser with swarms of agents, then open-sourced the repo.

There’s a whole skillset developing for being able to orchestrate and manage coding agents, with developers openly sharing their skills, sub-agents, and rules. (This is a great read on some best practices).

So the obvious question becomes, if building software is commoditized, what's your moat?

The value in Fintech SaaS

There was a recent viral trend for building a DocuSign clone, and I saw someone ask, “What the heck do their engineers even do if you can rebuild it in a few days.” The answer is edge cases. 

All the f*cking edge cases. 

In DocuSign’s case that’s things like:

• A built-in legal framework, providing legally binding eSignatures under laws like ESIGN, UETA, and eIDA

• Compliance with GDPR, eIDAS, and HIPAA gives enterprise buyers the warm fuzzies

• Adhering to Trust Authorities and the Adobe Approved Trust List (AATL) chain for cryptographic proof of signature/status.

• Integrations with Salesforce, Microsoft, and SAP, for 1.5 million enterprise customers, create high switching costs.

Who knew document signing was so complex? 

And as you know, dear reader, Fintech is a mobile app UI on top of the world's hot, chaotic lava full of legacy systems and compliance rules. Want to build an open banking API? Great, meet JP Morgan. They’ll expect a contract, your revenue, and your firstborn child. 

Plaid has data agreements with the large banks and is a CRA that lets you use users' cash flow data for underwriting. The power in data aggregation isn’t just raw data access its using it to price lending. Cash flow data (what a user spends) is so accurate

Want to build a compliance stack? Great, you’ll need a lot of proprietary and non-public data. Like originally sourced KYC documents, checks against government databases, industry databases (like EWS), user device & behavior data. You can buy all of that. But every time you buy another data set, you’re adding cost (COGS). 

Sardine* gives you a fraud and compliance toolset built for people who value performance. Sure, you could vibe-code similar features, but you might lack the expertise to build them that certain way that performs. When you’re buying these platforms, you’re also buying an opinion on what data is useful for preventing fraud or money laundering from an expert team.

The compounded advantage of founder-market-fit, network effects, and efficient scale that the existing Fintech SaaS companies have is non-trivial. 

The moat in Fintech SaaS is often not the distribution model (software hosted in the cloud). But in the expertise, data, and network effects those companies built and compounded to reach the point they’re at today. 

What if we just build it?

It used to be that you had to avoid side quests when building a Neobank. 

Now there’s less of a barrier.

Spending engineering time and effort on building the perfect compliance dashboard could be a distraction from the mission of saving customers time and money (in Ramp’s case), or helping cash stretch a little further (for Cash App). 

But now these companies have 10x more productive engineers, if owning the internal tool and self building it creates a better customer outcome there’s less of a barrier to building it. And, given they have some expertise and scale they can buy access to some of the data they’d need to do a good job at it. 

So yes you could unplug all your SaaS tools and build custom software.

But whether you should depends on if you can get enough benefit for the juice to be worth the squeeze. 

And some companies do choose to build solo. 

• Cash App has built their own Cash Score, using data from their customer base to price lending. 

• Brex built directly into payments networks which has enabled them to scale more quickly into new markets.

• Ramp is building all the AI tools - like a Policy agent.

Robinhood shipped $100m+ from 11 new product lines in the last 12 months. The speed at which these companies are able to ship new things is compounding.

What they choose to ship is the interesting signal. 

AI doesn’t eat Fintech SaaS - But it makes it much more competitive

The Fintech SaaS companies that survive aren't the ones who built great software. They're the ones who accumulated things agents can't generate: proprietary data, regulatory relationships, and the scar tissue from 1000 edge cases that only show up at scale.

If you're a Neobank deciding whether to build or buy, ask yourself: do I have the internal horsepower to burn cycles on identifying idiosyncratic user device behavior that could signal fraud? Or do I want my engineers shipping features customers actually see?

The moats are still there. 

But the drawbridge just got a lot faster.

ST.

4 Fintech Companies 💸

1. Nodu - MiCA aligned Stablecoin APIs for Fintech companies

Nodu allows regulated banks, PSPs, and money transmitters to quickly add stablecoin capability via its APIs. It aligns all payments and onboarding flows to AML/CFT and sanctions workflows with local laws and Europe’s MiCA regulation. 

🧠Every PSP and bank is looking for an infrastructure partner, and coverage in Europe is limited. Only recently, BVNK got its full SEPA readiness, and other infrastructure companies still have that on their roadmap. This is a team who’s built in payments before that knows Europe well (which is a complex regulatory market). This could be well timed.

2. Fido - Up to 10% APY with DeFi-powered savings

Fido is a simple mobile app that allows users to fund the wallet from their bank account, deposit funds, and earn up to 10% APY on funds. It also has gamified features to unlock further rewards.

🧠 The simplicity is the product. These ultra-simple UI’s are cheap to build on stablecoin infrastructure, and offer something that’s compelling. My question is how many of these apps will still be here in 5 years' time, especially if DeFi yields are correct.

3. Lighter - The Perps decentralized exchange on Ethereum.

Lighter is a new L2 (Layer 2) blockchain that uses Ethereum for its security. It enables ultra-high-frequency trading of “perps” (perpetual futures) at a speed professional traders expect. By building an app-specific ZK-rollup (a type of L2) they’re able to deliver much faster order matching and execution.

🧠 Traders love speed. We’re now in a place where on-chain performance is as good, and in some cases better than centralized exchanges. The “perp” is a very popular type of derivative, unlike an option or forward it has no expiry, so traders can get in and out of positions much faster and simpler. We’re not done at building new chain infrastructure yet. It’s still early.

4. Atomic Insights - The Money Mover for RIAs and Family Offices

Atomic Insights manages approval, compliance, and simplicity through a single dashboard for RIAs and Family Offices. It provides a real-time API to custodians, cash flow reporting, and integration to the CRM. 

🧠 Offices and RIAs have workflows like capital calls that is a pain that is not obvious unless you’ve lived it. It also has nice things like SSO support to Schwaab money. If you’re an RIA, this is a huge problem solver. 

Things to know 👀

1. Brex acquired by Capital One for $5.15bn

Capital One has agreed to acquire Fintech B2B spend management and CFO platform Brex for $5.15bn. Capital One, historically a consumer-lender, is acquiring one of the “big three” challenger corporate card and spend management platforms.

🧠 This is the single largest bank/fintech acquisition in history. Capital One gets a ready-made solution with “Brex-as-a-Service,” the horizontal solution Brex has doubled down on. Brex gets an acquirer that’s founder-led (Richard Fairbank), to operate separately, and Capital One will pour marketing dollars on them to grow.

🧠 Capital One gets a turnkey EU payment institution license, $13B in deposits, and clients like TikTok and Robinhood. They paid $5.15B for infrastructure that would've taken a decade to build.

🧠 What a comeback for Brex, who had a torrid pandemic:

• Bought a restaurant during COVID (???)

• Negative PR when they “abandoned” 10,000+ SMB customers

• Watched Ramp sprint past them while they restructured

🧠 What about the downround? Do the math. “$725m in capital raises above $5b, so assume that comes off the top of 1x liq pref. Approx $700m in debt (assuming not converted). That leaves $3.6b for equity holders.”

🧠 The lesson isn't "down rounds are bad." It's that liquidity beats valuation when the music stops.

🧠 Brex has become the “Adyen” of B2B Fintech focussed on large enterprises and ERP integration. This makes the Brex platform uniquely simple to integrate into Capital One’s go-to-market. I’m generally dubious about companies acquiring start-ups, but this one could have legs.

🧠 There’s a fascinating contrast with Ramp.

• Ramp's last valuation: $32bn.

• Same market. Often considered peers.

• Ramp is growing faster and took less time to do it.

• Ramp's founders previously sold Paribus to... Capital One.

You couldn't write this in fiction.

 🧠 How is Fifth Third feeling about this acquisition? Fifth Third signed a partnership with Brex for commercial cards in early December. I imagine their deal will be maintained, but does Capital One want to be in the business of serving its competitors that way?

🧠 Where does the Discover network fit in with all of this? The Diners Club commercial card did $622bn of transaction volume in 2024, but for M&A, maybe that’s a red herring.

🧠 For the team and founders, this could be a great deal. $12.3bn → $5.15bn sounds like a loss. But $3.6bn of upside is meaningful for folks like Ribbit, who led a $7m Series A. For the founders and team, this is a great outcome. Brex once looked troubled and overvalued. Now it has executed one of the greatest comebacks and exits in Fintech history.

2. Australian Financial Regulator investigates Airwallex after AML concerns

AUSTRAC the Australian financial crimes agency,” is investigating Airwallex and has forced it to engage a special auditor after becoming concerned that the money transfer platform was being used to fund child sexual abuse.” According to AFR, the regulator was concerned the company “was not properly monitoring payments it was processing.” The news follows Airwallex's recent funding and planned accounts. The agency went on to say they had concerns of “money mules” accounts in real names used by criminals to funnel money abroad. The company said it expects a special audit to validate its existing practices.

🧠 Money mules are a massive issue for everyone, everywhere. If you speak to any bank or regulator, it's become an incredibly fast-growing problem for them all.

🧠 Solving it is non-trivial; a legit customer, with a real identity, signs up and starts moving money. How is the company opening the account supposed to know that the person has been compromised?

🧠 The most common controls are KYC and Transaction Monitoring. Getting more data about the customer, their previous bank accounts, government, and consortia database checks, etc., can help validate if someone really can afford to move $1000s of dollars a month offshore on a smaller salary. Or if that person has a history of opening lots of new accounts.

🧠 Cross-border makes this 100x harder and riskier. If you’re moving funds around lots of countries between companies and individuals, now you need those company registries, and the history of the owners. In every single market. Often, where data quality is poor.

🧠 Every company does this to some extent but it’s hard and nobody does it perfectly. The regulators often question if enough is being done to continually check, improve and monitor.

🧠 For some added drama famed investor Keith Rabois has publicly called out Airwallex as “Chinese engineering and legal apparatus with a global facade” something the company streneously denied. I often wonder, how can you operate a cross-border business, especially into and out of APAC, without getting dragged into Geopolitics these days?

🧠 Time will tell what happens next. Airwallex is still growing incredibly quickly, shipping new products, and is on a path to an IPO. Will this be a bump in the road or a major obstacle?

🧠 This story, and Kontigo recently, are strong reminders that compliance is not optional. Moving money is hard. And getting into Fintech is easy. Staying there is the real battle.

2. Trump calls on Congress to implement the 10% APR cap on credit cards

From Finextra: “President Donald Trump says he will ask Congress to implement a one-year 10% cap on credit card interest rates in a move that JPMorgan chief Jamie Dimon says would spell “economic disaster” for the US.”

🧠 Jamie Dimon says capping rates, “would remove credit from 80% of Americans, and that is their back-up credit." 

🧠 Not to call out the GOAT here, but he’s sort of right and sort of wrong.  

🧠 He’s right because obviously a cap on APR just means lending becomes uneconomic and consumers no longer get credit.

🧠 He’s wrong because we live in a world where BNPL exists, and Cash App and Chime are innovating on what’s possible with cash flow underwritten credit.

🧠 The fact is credit cards are a cash cow business for the big banks. Often at the expense of the lower-income consumer population.

🧠 I doubt Congress will go for this, and the bank lobby is using every platform at Davos to call it what it is, stupid and unworkable. But my point is, credit cards have a major competitor in BNPL and Neobanks. Today they’re still small, but moves to expand credit affordability should (and will) come through competition, and if we legislate it should be to encourage that.

Other things worth your attention

• Revolut added account-to-account (pay by bank) to Google’s A2P Agentic Commerce Protocol and rolled it out in Revolut Pay. You should pay attention to this because Revolut Pay is quietly becoming a force in online payments. If agents will choose to buy from places that support them. They can also choose to pay by bank and select Revolut Pay if they see it available, which means more sales for merchants and more volume for Revolut. That’s a potential innovation dividend.

Good Reads 📚

1. Kontingo the “Sanctions Evasion” startup

Kontingo, the stablecoin wallet start-up has enabled transactions to sanctioned entities such as Banco de Venezuela in Venezuela, with the founder having taken screenshots of off-ramping to that bank. This was followed by announcing the ability for “Venezuelans to open free “virtual” accounts at JPMorgan Chase in order to move money.”

🧠 Jason kept the receipts. This piece is full of screenshots of the above and much more and stands as a stark warning to everyone in the stablecoin space.

🧠 Kontigo seems to think threats are the answer to their negative PR. In reality, it just further damages their credibility. Plenty of founders have been on the receiving end of it and are still building great businesses. They did so by also addressing the issues, not threatening and attacking those who raised them.

Tweets of the week 🕊

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