3 FINTECH NEWS STORIES
#1: Congress is Actually Trying to Do Something!
What happened?
A group of Senators and Representatives in the U.S. Congress are proposing a new law:
A bipartisan group of members unveiled legislation that would direct federal financial regulators to create “regulatory sandboxes” for financial firms to experiment with artificial intelligence.
Led by Sens. Mike Rounds (R-S.D.) and Martin Heinrich (D-N.M.) plus Reps. French Hill (R-Ark.) and Ritchie Torres (D-N.Y.), the bill is titled the Unleashing AI Innovation in Financial Services Act.
This sandbox bill from Rounds, Heinrich, Hill and Torres would allow firms to apply and “experiment with AI test projects without unnecessary or unduly burdensome regulation or expectation of retroactive enforcement actions.”
That application would let firms request that compliance with specific regulations be “waived or modified.” Companies could also propose an “alternative method” for complying with existing regulations.
Companies would need to demonstrate their project wouldn’t present a “systemic risk” to the U.S. financial system and comply with anti-money laundering laws.
So what?
A bipartisan, bicameral bill on banking and AI? In 2024?
Am I the one hallucinating?!?
In all seriousness, I love this idea, and I applaud Rounds, Heinrich, Hill, and Torres for proposing it.
We haven’t really embraced the idea of regulatory sandboxes at the federal level here in the U.S., though other countries like the UK have, with mixed results.
The effectiveness of regulatory sandboxes depends heavily on the operational details. Is participating in the sandbox significantly less risky and expensive than the more traditional go-to-market strategies? Does successful participation in a sandbox provide a government imprimatur that makes it easier for graduated participants to win in the market? Do the limitations and supervision imposed by the sandbox effectively limit systemic risk and ensure that consumers understand exactly what they’re signing up for?
Getting these details right is difficult, and it’s too early to say if this legislation (and its implementation by federal banking regulators) would have the desired effect.
Having said that, I think we need to try something like this.
Generative AI has tremendous potential to reshape banking in ways both beneficial and detrimental to consumers. Regardless of what regulators say or do, companies will experiment with generative AI, and not just in a “we only use LLMs to improve the efficiency of mundane back-office tasks” kinda way. They will unleash generative AI into customer-facing use cases, including loan underwriting. It is inevitable.
It is also inevitable that criminals will leverage generative AI to steal people’s money (indeed, this is already occurring).
I would prefer that we create a space for responsible actors to innovate in these areas — for the benefit of customers and the utter ruination of fraudsters — under controlled conditions, with consumers who fully understand and accept the risks that they are taking.
#2: AI for E-Invoicing
What happened?
A fintech infrastructure company took off its cloak of invisibility and announced a seed round:
Monto … today exited stealth with $9 million in seed funding to bring order to B2B payments collection chaos. Monto enables B2B finance teams to seamlessly get paid from any AP portal used by their enterprise customers.
So what?
Staying on the theme of AI in financial services, this is a product that I am very intrigued by. Here are a few additional details on the opportunity Monto is going after:
Regulations worldwide are pushing businesses to establish their own e-invoicing platforms. According to Deloitte, companies worldwide will adopt e-invoicing, whether mandated by governments in the case of Europe, or from market leaders supported by the government. In this new B2B world, suppliers must integrate with each customer’s payment software, each with its own complex logins and workflows, creating endless chaos for payment collection teams that can severely impact cash flow and working capital.
Monto has created smart connections between financial systems, primarily between ERPs and AP portals. Its AI-based platform learns each customer’s invoicing requirements to ensure a seamless payment flow for the supplier to significantly reduce manual workload, mitigate the risk of late payments and improve cash flow management.
This is a pattern that we have seen repeatedly in financial services over the last 20 years and one that I think we will continue to see for the foreseeable future — the digitization of different industries creates a myriad of new customer-facing portals and developer APIs, which creates new customer-permissioned read/write opportunities for software developers.
Historically, the challenge was that it took a lot of work to build and maintain scrapers or other one-off integrations to interface with these various customer-facing portals or proprietary APIs. I am intrigued by the potential of generative AI to build and (perhaps more importantly) monitor and maintain these custom integrations in a far more automated manner.
Of course, Monto and anyone else who adopts this playbook will face the challenge that their scraper bots and custom integrations no longer act as a competitive moat against incumbents or other startups. But such is life in the age of AI.
#3: In The End, There Can Only Be One
What happened?
Pagaya is acquiring Theorem:
The combination brings Theorem’s consumer credit funds as well as its engineering and data science expertise under the Pagaya umbrella. This strategic transaction solidifies Pagaya’s overall mission to serve its partners, their customers and investors by expanding access to credit across the lending ecosystem.
In aggregate, Pagaya’s fund management business is expected to grow to more than $3 billion of capital in investment vehicles separate from and incremental to the Company’s market-leading securitization program, in line with the Company’s financial strategy to diversify its funding sources and enhance its capital efficiency. In addition to its existing investment opportunities, investors in Theorem’s credit funds will now have access to credit assets generated by Pagaya’s network of 30 of the top lenders in the US, including over $180 billion of application volume per quarter.
So what?
If you need a refresher on what Pagaya and Theorem do, I recommend reading this essay that I wrote a few months ago.
The explanation, in brief, is that Pagaya and Theorem provide embedded second-look financing, in which a lender can extend their credit box lower than they would otherwise feel comfortable, by integrating with Pagaya or Theorem’s real-time loan underwriting systems. This allows the lender to approve more applicants and to service and cross-sell more customers, while, in the background, passing the risky loans on to Pagaya or Theorem, which have built their own proprietary networks of yield-hungry secondary market investors.
It’s a tremendously compelling option for lenders (say yes more with no additional risk!), but, from my perspective, it’s not a sustainable business model. Theorem was founded in 2014. Pagaya was founded in 2018. They came of age in a low-rate environment, in which it made sense to build the infrastructure to enable risk-on investors to find yield. However, as interest rates have risen, Pagaya and Theorem have had to reach deeper into near-prime and subprime lending tiers (subprime auto, in particular, has been a big area of focus for Pagaya) in order to continue generating above-market yield for investors.
Now the embedded second-look financing space is consolidating down from two providers to one, with the acquiring company talking explicitly about how the acquisition will strengthen the investment side of its business (“Pagaya’s fund management business is expected to grow to more than $3 billion of capital in investment vehicles”).
I have a bad feeling about this.
2 FINTECH CONTENT RECOMMENDATIONS
#1: The Next Wave Of AI In Banking: Intelligent, Autonomous Agents (by Ron Shevlin, Forbes) 📚
An excellent article from Ron on the impact that autonomous, intelligent agents (powered by LLMs) will have on banking. My personal favorite example is the “ReFi Robot” idea initially proposed by the folks over at a16z — helps consumers continuously find and apply for loans to lower the cost of their existing debts.
When this takes off at scale in financial services, a lot is going to change.
#2: How to make money selling money (by Matt Brown) 📚
I plugged this essay in my piece last week on the necessity of being acquired by your customers rather than acquiring them, but I wanted to recommend it again here.
It’s outstanding, and very much worth a read!
1 QUESTION TO PONDER
What’s the difference between a ledger and a core system?
Everyone in the BaaS middleware space is pivoting and referring to themselves as a core provider (even if nothing substantial with their model or tech is changing), so I’m very curious about the answer to this question.
