🧠 How do we fix issues that led to Synapse/Evolve?

Welcome to Fintech Brainfood, the weekly deep dive into Fintech news, events, and analysis. New here? You can join by clicking below

Hey Fintech Nerds 👋

Generative AI is having a meaningful impact on financial services today. Klarna says AI investment has reduced Opex by 11%. There’s a lot of hype and noise, but there’s also real progress.

Every Neobank will be a Telco; Nubank and Revolut now do eSIMs, and they’re becoming wallets that do everything else. The future of Fintech is wallet wars.

Embedded Finance is still a train wreck, but we can fix it (that’s your Rant this week).

PS. Episode 2 of the Fintech Brainfood Podcast is out now. I interviewed Pedro Francheschi, the Co-CEO of Brex, and the contrast with Ramp is fascinating.

• Ramp, ships daily, leverages partners and product is the center of gravity for decisions

• Brex, ships quarterly, vertically integrates, and Pedro is the arbiter of the roadmap

Available on your favorite podcast client now

Here's this week's Brainfood in summary

📣 Rant: How do Fix The issues in BaaS that led to Synapse/Evolve?

💸 4 Fintech Companies:

1. Tabs - LLM-powered Billing and Accounts Receivable.

2. Parcha - KYC and AML Investigations AI Agent

3. Octa - Accounts Receivable Automation

4. Teal - Embedded Accounting as a Service

👀 Things to Know:

1. Nubank is launching eSIMs

2. Klarna says it reduced Operating Expenses by 11% - driven by AI

3. CLEAR and Public partner to streamline onboarding

📚 Good Read: A few reasons to be bullish on PayPal

Weekly Rant 📣

Can we fix BaaS with better standards or with a FINRA-like entity?

We've seen some of worst case scenarios: customers locked out of their accounts, unable to pay bills, and a giant public spat between banks and intermediaries who are supposed to keep all of that safe.

Up to 10m consumers could be impacted, and nobody seems to know whose problem it is.

This is the fallout from a high-profile court case for the insolvency of BaaS provider Synapse, its former partner bank Evolve. While a lot of blame is passed around, the fundamental issue comes down to one thing.

Nobody knows the state of the ledger.

This has left the bankruptcy court to try to unpick the mess involving Synapse, Evolve, and countless other Fintech companies.

The nightmare scenario has played out because we lack standards, self-regulation, or some level of good governance enforcement.

We need to do better. We can do better.

Why does nobody know where the deposits are?

Nobody can say with confidence which customer has which deposits. As Jason Mikula describes here

First, the banks dispute the Synapse record.

The Fintech dispute Synapse's record, too.

For its part, Synapse says the issues began when Mercury (one of the largest Fintech banks and former Synapse client) decided to go directly to its partners like Evolve.

Everyone has different payment, ledger, and accounting data.

Nobody can agree on who's owed what. (Although it seems the ledger issues run deep).

This is going to take forensic accounting to fix.

Here's a few things that led us here and where we could go next.

1. Compliance was loosely enforced by sponsor banks. With many documented examples of consent orders and failings from regulators, we've seen failures in disclosures, UDAAP, and countless other compliance issues.

2. Third-party oversight was limited. Banks can and do use third parties, but they have to ensure those companies are resilient and can maintain the right records.

3. It was inevitable that Fintech programs would look to go directly to banks as they scaled. Cutting out the middleman improves unit economics and is what the regulators are pushing for.

4. By taking on compliance and ledgering, BaaS providers took on a responsibility they couldn't always live up to. The benefit was time-to-market for Fintech programs. The drawback was neither the program nor bank prioritized building a golden customer view of deposits.

5. Therefore, nobody can agree on the state of the ledger. Making it impossible for the court to figure out how to make customers whole.

The simple answer is to say, "These are all the bank's customers," and enforce that model.

That wouldn't fix the issue of nobody knowing the state of the ledger if one Fintech company can have multiple bank partners.

Where do regulators fit in all of this?

Great question.

In theory, there is a law and regulator responsible for all of this in the US. The problem is, to quote Michele Alt.

I think there's a far better solution that protects consumers.

The European Electronic Money Institution (EMI) license helps solve this (and several other) issues.

Lessons from the Electronic Money Institution

An Electronic Money Institution (EMI) license exists nationally (e.g., in The UK). It occupies a space between the FBO account, UDAAP, FDIC, CFPB Wallet rules, and the Money Transmission Licence.

They require:

1. A BSA Officer (Money Laundering Reporting Officer) and CEO who's based in-country and is viewed by the regulator as having suitable experience to carry out their role

2. Safeguarding of customer funds, either by individually segregating in ring-fenced client (omnibus) accounts at a bank (or in the UK at the central bank).

3. AML/KYC, IT Security and internal risk policies must be submitted as part of the licence and reviewed by the regulator

4. Are subject to consumer duty (UDAAP++) requirements to ensure good outcomes for consumers, avoiding dark patterns, clear communications, and support for vulnerable groups

5. Subject to full disclosures and financial promotions rules (e.g. FDIC-style deposit disclosures in marketing)

To achieve this, you'd need to combine the responsibilities of at least 7 national and local state regulators.

Regulatory simplification is a great goal, but it's often like sedimentary rock. New regulators get added but rarely removed.

The OCC did try to introduce a National Fintech Charter in 2018/2019, but it was subject to litigation from state banking agencies and the Conference of State Banking Supervisors that argued the national regulator doesn’t have chartering authority from Congress to issue these accounts (the history of which is fascinating).

So I doubt we'll get a federal MTL/EMI any time soon, but maybe there’s another option.

What if the industry regulated itself?

If we can't have an EMI can we have a FINRA for BaaS?

The Financial Industry Regulatory Authority (FINRA) is a private organization that acts as an industry regulator for brokerage firms and exchange markets. As a self-regulatory organization (SRO), it has delegated authority from the SEC to supervise its members.

Today, FINRA it reports on practices such as fair pricing, financial crimes, and cybersecurity and examines its members for compliance with these rules.

Perhaps the Banking-as-a-Service and embedded finance industry needs a coalition of willing banks, Fintech programs and providers to come together to raise the bar.

This is a wild idea (credit Kiah Haslett who I first saw mention it)..

The problem here is again, the federal and state regulators would probably end up in a jurisdictional bum fight unless congress stepped in.

So behind door number #3

What if we had a SOC2 for BaaS?

SOC2 is the cybersecurity compliance requirement adopted by the American Institute of Certified Public Accountants (AICPA). Companies are required to prove they have implemented security controls (vs prove they have a process in place).

Independent auditors will then audit the proof pack an organization’s security team puts together. This voluntary standard signals to customers, partners, and regulators that your organization takes cybersecurity seriously.

Now imagine this for ledgering, 3rd party oversight, BSA/AML, secure data sharing, risk assessments, partner onboarding. The list goes on.

There’s even a great starting point. The folks at Alloy Labs put a standard together.

This is just one idea.

What would you do to fix BaaS?

There are probably 100 other good ideas that could go here.

I'm working with the folks at the Association for Innovation in Regulation, several community banks, and providers to kick off a Techsprint in Washington to look at this and multiple other possible ways forward for BaaS.

Because the answer can't just be more bad news.

Spoiler: This techsprint would need funding 💸

If you're keen to support, hit reply, and I'll connect you with the folks at AIR.

I bet you didn't have—let's build BaaS standards or Self Regulating Organizations on your bingo card for 2024—but life comes at you fast.

ST.

4 Fintech Companies 💸

1. Tabs - LLM-powered Billing and Accounts Receivable.

Tabs is an all-in-one Accounts Receivable platform that manages the process from contract to cash. It is unique in that it can use LLMs to read a contract and extract billing terms automatically. Tabs also automate invoicing and will manage follow-ups for late payments to ensure companies get paid faster.

🧠 FinOps as a service. Billing and payment collection is complicated. Managing a handful of enterprise clients' monthly billing can easily take a person's time. Now, imagine you grow to a few hundred companies. Being able to extract that, manage invoicing, and chase collections? Amazing. I want this in the UK.

2. Parcha - KYC and AML Investigations AI Agent

Parcha helps review businesses and individuals during KYC and KYB onboarding before a manual review. It will evaluate incorporation documents, Government IDs, or Tax IDs and perform a "pre-case" management check. If more information is required from a customer, it can also prepare email correspondence before the customer is onboarded.

🧠 The nature of AML investigations is looking through mountains of documents to create a case. Having an AI agent eliminate 90% of that work and drive most of the way to a decision is a huge efficiency gain. That will still need human oversight, but fundamentally it allows financial institutions and Fintech companies to say no faster and/or onboard more customers.

3. Octa - Accounts Receivable Automation

Octa allows companies to track, chase, and collect payments against outstanding invoices. It helps reconcile all payments into an accounting system and offers your customers flexible payment plans. For example, a company could offer discounts for early payment.

🧠 A lot of thought has gone into the detail here. The product is trying to do everything to create an invoice that is likely to be paid and help companies incentivize on-time payments. 2024 is the year of accounts receivable. FinOps teams everywhere rejoice!

4. Teal - Embedded Accounting as a Service

Teal allows platforms and vertical SaaS companies to sell accounting software to their customers. Companies can offer their clients white-labeled bookkeeping, tax filing, and CFO support within a single interface.

🧠 The internet (and Fintech) is a fractal. If you zoom in on every department in a digital business, you'll find 100s of SaaS companies doing the heavy lifting. The same is true for payments and financial services. These tasks naturally belong together, and making them embeddable is inevitable. It makes you wonder what the center of gravity is for a financial operating system? And does that vary by vertical from accounting to banking to something else?

Things to know 👀

1. Nubank is launching eSIMs

Every Neobank will be a Mobile Network? Following Revolut's launch of eSIM functionality, Nubank will give mobile service via eSIMs to travellers in more than 40 countries. The news follows Brazil's National Telecommunications Agency greenlighting Nubank becoming a full MVNO.

What's an eSIM? The electronic SIM card, gives much lower cost data, calls, and text when traveling and can be instantly provided to your mobile phone)

🧠 They wouldn't launch this feature unless customers wanted it. There's a clear demand for it, and it helps the Neobanks attract a more profitable traveler customer segment. This is especially useful for Nubank, which has been stronger in subprime historically.

🧠 eSIMs are a great way to convert users into subscribers. Affuluent customers travel more, and data roaming fees are the kind of speed bump and niggle that people often accept but rarely fix. Bundling it into your Neobank is neat. It brings the Neobank closer to travel, and it brings you the user, closer to paying a subscription with them.

🧠 How long until every bank tries to offer this feature? The classic "premium account" with mobile phone insurance has passed its prime. On the other hand, eSIMs are a great way to solve roaming fees and are of great value. Most people don't use specialist eSIM apps so why wouldn't it be bundled right next to where you manage your money?

2. Klarna says it reduced Operating Expenses by 11% - driven by AI

Klarna saw a 29% increase in revenues YoY driven by its US expansion while reducing operational expenses (Opex) by 11%, which was attributed to its investments in AI. The company said 90% of employees now use AI in their daily work, 4m customers have used the new AI service

🧠 The Klarna “we’re ready for IPO” tour rolls on. Bloomberg reported a possible $20bn IPO later in the year. The constant AI, cost reduction, and new product launches PR are in service of that to some extent.

🧠 You can’t argue with results. These are good numbers. BNPL is a fantastic wedge, but they deliver much more of a “shopping experience” for their high-value customers. They’re moving into branded checkouts and loyalty. This is an absolute play to be the consumer’s wallet and center of gravity for shopping.

🧠 The dog whistle of “Look how efficient we are with AI, but look at the job losses” is savvy, but I see it. Every time they announce a new feature like how AI is doing 90% of their marketing images, they also “acknowledge the wider societal impacts.” It’s OMG, we’re so efficient, but also, oh no, but think of society. Big companies need to sound responsible on AI because there are so many AI-doomers out there, but I can’t help wanting to vomit every time I read that sentence.

🧠 I was cynical about GenAI chatbots but actually had a great experience recently. LLMs are implemented poorly, hallucinate, and never quite do what you want them to. Well harnessed, they’re magic. I had to remove a custom domain from the substack to get it here to Beehiiv, and honestly, that chatbot was amazing. It figured out what I wanted and made it happen.

3. CLEAR and Public partner to streamline onboarding

The CLEAR service, best known for digital and biometric identity at airports and sports stadiums, has partnered with Public .com, the brokerage, to bring onboarding down to 1 minute. CLEAR members that pay for the service will also receive a $100 credit on their public brokerage account (or $50 for members who have used the service outside the airport.)

🧠 This is the future of digital identity in the Anglosphere. Networks like CLEAR will be implemented in physical spaces and support digital/Fintech onboarding use cases afterward.

🧠 The competitor is storing identity in digital wallets like Apple Wallet. States have rolled out the ability to store licenses in Apple (and now, Samsung) wallets. These services will co-exist and use standards published by the US Government.

🧠 It's a clever customer acquisition for both sides. First, it gets them both a press release, second it opens a pool of 22m new users to public, and third, it gives a reason for people to open or upgrade their CLEAR account.

Good Reads 📚

1. A few reasons to be bullish on PayPal

Jvgenis, your new favorite Fintech analyst, nails it. "PayPal stock is down 42% over 5 years, while the NASDAQ was up 81%, and PayPal is losing share to Apple and Shop Pay". It's actually doing well internationally and with enterprise, but can the original branded checkout re-ignite that business?

Great stats in here. Branded checkout is its most profitable business line by far (unbranded 0.3% take rate, vs 2.1% on branded), US is 64% of TPV, international is 42%.

🧠 Unbranded checkout Braintree is doing well, but not as well as Adyen or Stripe. Jvgenis points out that Braintree uses Fiserv for processing, so it doesn't benefit from the same margins that Adyen or Stripe enjoy, who own the processing infrastructure in-house. Despite that, Braintree is growing and taking market share from incumbent processors in the US.

Here's what I'd do

🧠 Step 1: Bring processing in-house is a no-brainer. This would be a large, multi-year engineering commitment that the market will initially hate. A new CEO must explain this to the market as healthy for long-term gross margins. Braintree is #3 in a 3-horse race, taking e-commerce processing market share from incumbents.

🧠 Step 2: Push hard at the branded checkout experience. Study Shopify and Apple Pay closely. They are astonishingly good products. It's about so much more than clicking "pay." It's about storing address details, managing delivery and logistics, etc. (Branded on in-house processing would have god-tier margins). PayPal is doing interesting things with receipts, and returns are a huge merchant issue. Solve that.

🧠 Step 3: Double down on international growth. Shop Pay and Apple Pay aren't nearly as dominant internationally as they are domestically. They appear when buying from a Shopify business or a US-based platform company. PayPal also has some growth in APAC, an area Adyen or Stripe are having some early joy partnering to win.

🧠 Step 4: Build a pay-by-bank story. Pay-by-bank is the next big emerging rail. Using ACH or domestic payments for Airbnb to Uber to X subscriptions is becoming a default in the US and Europe. Where is PayPal on this?

Tweets of the week 🕊

That's it, folks. 👋

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

(1) All content and views expressed here are 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 see it as gospel—strong opinions weakly held 

(5) Citations may be missing, and I do 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