After four years of development (and 10 years since its announcement), the Fed launched FedNow this past July with 41 banks, including JPMorgan Chase, BNY Mellon, and U.S. Bancorp, and 15 service providers approved to use its services. A real-time payments mechanism for individuals and businesses in the U.S., FedNow is the country’s long-awaited version of a faster payments service that has long been available in the U.K., Brazil, and India, among other countries. However, despite its launch, much on the operations side still needs to be figured out.
First off, going live doesn’t mean that consumers can use the product yet. Not only do financial institutions (FIs) have to decide which use cases they want to prioritize (e.g., billpay, peer to peer) first, but as we’ve discussed before, any financial institution that wants to work with FedNow will have to consider if it wants to invest in the technology required to connect to and maintain a connection to the service.
Fraud and risk controls are also major concerns. Each FI on FedNow will be responsible for creating their own end-user interface and implementing most of the security around the faster payment transactions. The Fed has said it will launch some controls, but the responsibility sits squarely with the FIs. To help temper the risks, many banks will start with “receive only” transactions, even though send and receive are both currently possible through FedNow. Transactions will also start with a $100,000 limit, which banks can choose to increase up to $500,000.
Pricing is also still unclear. For 2023, the Fed will waive the participation fees for banks, though they will have to pay a monthly $25 fee for each routing transit number enrolled to receive credit transfers from the FedNow service starting in 2024. Reserve banks, meanwhile, will introduce payment and transfer liquidity management fees in the future. However, it’s unclear how this existing pricing will translate into costs to consumers, especially vs. ACH.
Finally, the Fed still needs to think about interoperability with existing payment methods, that is, how customers from different banks—that have access to different, faster payment schemes, such as those on the existing RTP (real-time payments) network—can send money to each other.
In the long run, the competition spurred by FedNow will be valuable for consumers. RTP, Zelle, Venmo, the card networks, and others all already offer faster payments, but with more competition, end users will benefit from improved efficiency and infrastructure across the board, assuming it’s safe, fast, convenient, and affordable to do so. However, only 1.2% of all payments in the U.S. are currently sent via faster payments according to an ACI Worldwide/GlobalData report. There are also nearly 10,000 financial institutions in the U.S., so there’s still a ways to go in enrolling participants on FedNow. The Fed is particularly keen to attract to the service smaller U.S. banks, who have traditionally not connected to the existing RTP network because it is operated by The Clearing House, a company owned by their larger bank rivals.
It will be interesting to see, as the FIs adopt the technology to operate on these new rails, if the various faster payments providers segment themselves based on use case, size of payment, domestic vs. cross border, and more. We’ll also be following if use cases gravitate toward faster payments (e.g., RTP ended up attracting more consumer and B2C use cases like wage advance, instead of just the anticipated B2B use cases).
One of the most exciting advances of generative AI is the fact that LLMs, like GPT-4, can process and output both text and images. A result of this technical breakthrough is the potential for “agents” to execute actions on your behalf. My partners Anish Acharya and Olivia Moore recently dove deep on the ramifications of this technology on consumer financial services, predicting the rise of consumer robot process automation (RPA) that can finally deliver on the vision for “self-driving money.” In this future, the cost (and friction) of applying for financial services effectively declines to zero, enabling an “AI agent” to constantly scour the financial universe (and importantly, apply on your behalf) for the best offers/rates/terms “hands free.”
While this innovation could be hugely beneficial for consumers—effectively democratizing financial planning and wealth management for the masses—it may have an equally profound (and opposing) impact on the manufacturers of these financial products. Customer loyalty is likely to decline and the velocity with which deposits, loans, and investment accounts move across institutions could cause significant balance sheet and liquidity issues. It may also significantly stress operational P&L at financial institutions, as the volume of (programmatic) applications is likely to skyrocket, which will tax their underwriting and operational staff.
In order to effectively navigate this future, financial institutions will need new AI-native tools to more accurately assess “actual” transactional intent vs. this new breed of “programmatic rate shopping.” Similarly, a new AI tool kit will be required to combat new vectors of fraud, including identity spoofing with new voice models, as well as falsely generated financial statements and other documents. Lastly, new methods for processing normal applications more efficiently will be required as well.
We’re excited about this new category of AI-native financial software products and have been spending time with many of the country’s largest financial institutions to better understand their needs, as well as the opportunities and challenges they anticipate. If you’re building in this space, we’d love to chat!
Robotic Process Automation (RPA) companies help enterprises automate manual tasks. UIPath (whose market cap exceeds $8 billion) alone has deployed thousands of bots within financial services to help automate tasks like account opening, compliance, data extraction, and migration across many functions.
Despite the availability of RPA, banks still infamously employ tens of thousands of people to do manual tasks. Why? This is partly because banks haven’t digitized all possible processes, but a bigger reason might be that we are running into the limitations of what RPA can do. This is where generative AI comes in.
While RPA excels in automating repetitive, rule-based tasks, AI, particularly GenAI, excels at processing unstructured data and making decisions based on more complex inputs.
For example, let’s look at how a bank uses RPA to handle the KYC (Know Your Customer) process when a new customer creates an account. The series of bots would:
However, despite these abilities, these bots also have many limitations including:
A GenAI agent, on the other hand, uses advanced optical character recognition (OCR) to read and understand a broader variety of documents, even if they deviate from the standard template. It can also process more complex information, such as self-declared occupations or the purpose of the account.
Additionally, by analyzing global news, databases, and other unstructured resources, GenAI can identify potential risks related to a customer based on emerging global events or updates. It can also automatically categorize and prioritize applications, so high-quality, low-risk applications might be fast-tracked, while riskier ones are subjected to more in-depth scrutiny. And that’s just the beginning.
RPA companies are already “adding AI.” Yet there are more billion-dollar companies to be built by teams who deeply understand the hundreds (thousands?) of disparate tasks still to be automated within banks, insurance companies, etc., how buyers and procurement departments work, and most importantly, how to apply the latest AI infrastructure toward solving these problems.
Let’s imagine you are building a new insurtech product. You will need to find customers, build (or leverage) a technology platform to operate the business, employ an underwriter to help evaluate which risks to cover and how much to charge, and finally, work with a licensed insurance carrier and reinsurer to cover the risk you are writing. As the distributor and underwriter, the new insurtech, along with its insurance carrier and the reinsurer, will be regulated by state and federal officials. Each party has to ensure they are running the right procedures to guarantee their customers are protected from compliance issues such as fraud. Each party is also dependent on the other to comply with their own relevant regulations.
If your new business grows quickly, you might need to find additional reinsurance capacity. Vesttoo, an insurance startup that connects the capital markets (notably non-insurance-specific investors) to reinsurance deals via a structure called “insurance-linked securities,” was built to help with that issue, though it’s recently run into some complications.
By creating a marketplace where it could price insurance risk for investors who are unfamiliar with the process, Vesttoo increased the amount of investor capacity available to insurance carriers, traditional managing general agents (MGAs), and startup MGAs who need it. By providing additional investor capital into a “hard” (insurance speak for capacity-constrained) insurance market, Vesttoo’s business scaled well; it was valued at $1 billion as recently as last October. However, a recent compliance issue dramatically impacted the company.
When Vesttoo was facilitating a new reinsurance contract, each contract needed to post a letter-of-credit (LOC), which functions as collateral on the deal. These LOCs are sourced from banks, and in Vesttoo’s case, multiple billions were sourced through China Construction Bank. While it’s not yet clear if the subsequent breakdown was due to a compliance or due diligence failure, what is clear is that there are billions of dollars in fraudulent LOCs associated with the Vesttoo platform, leaving many MGAs, insurers, and reinsurers with large capacity holes that need to be filled immediately. Since the news about the fraud issues broke, Vesttoo has filed for bankruptcy, replaced its CEO, and laid off 75% of the company. The FBI is also investigating the fake LOCs and Aon has taken legal action for $137 million in losses.
This sad result has had a painful financial impact on all parties involved and accelerated a general flight to quality that is making a hard capacity market even harder—especially for startups. The entire affair is also a depressing reminder of what can go wrong when compliance processes fail to identify risks, whether they are first-party or third-party risks, and why it’s of paramount importance to have the right compliance teams and processes in place as a startup scales.
Peer-to-peer (P2P) electronic payments continue to represent a large opportunity in Mexico, where 90% of transactions below $500 pesos are still in cash. After 2018’s rollout of CoDi failed to gain traction, the Central Bank of Mexico is now in the process of launching a new initiative called DiMo (“Dinero Movil”). We believe that this initiative might have the right ingredients to crack mass adoption of P2P payments.
DiMo will launch in different stages later this year first as a P2P network, before it offers peer-to-merchant services. It will be completely free and users will be able to associate their phone number with their account, much like Zelle users in the U.S. do. Additionally, in a country where the World Bank estimates more than 50% of the population doesn’t have a bank account, DiMo will remove much of the friction in opening one by providing information to consumers for a variety of banks and licensed fintech options.
DiMo has the opportunity to be for Mexico what Pix is for Brazil. If it gains traction, we believe that it will spur a new wave of innovation and we think we have an idea where it will start. Please reach out to us with your thoughts about DiMo and stay tuned for more.
The Consumer Financial Protection Bureau (CFPB), the U.S. Department of Health and Human Services (HHS), and the U.S. Department of Treasury recently launched an inquiry into the various forms of patient finance that are available to consumers at or before the point of sale in healthcare settings; these primarily include medical credit cards and “buy now, pay later” installment loans. The inquiry will seek information about the prevalence of these products, how and when patients use them, and the incentives that care providers may have to offer them.
From the press release announcing this inquiry, it’s clear the CFPB’s primary concern is that these products could ultimately drive up healthcare costs and add to the already staggering level of medical debt in the U.S.
A separate CFPB research report released in May goes into more specifics about its concerns, which include:
While the above concerns represent the initial findings of CFPB research, it’s important to note that this inquiry is still in process. We’ll be watching this story closely as it develops, as patient finance has been a hugely popular area of interest for builders at the intersection of healthcare and fintech. Any further regulation to patient finance could have serious implications, such as shrinking existing portfolios for existing businesses, significantly impacting new loan originations, or prompting incumbents/payvidors to potentially bake in financing schemes into their offerings. Meanwhile, there is also an opportunity to rethink the product experience and focus on delivery through non-provider distribution channels to circumnavigate the issue of potentially misaligned incentives.
Our hope is that rather than stymying innovation in a space that is capable of providing crucial access to various forms of medical care, the CFPB inquiry pushes both incumbents and startups to keep at it, while refining their product and pricing strategies to be as patient friendly as possible.
A dynamic new wave of digital-first companies has emerged, driven by a significant decrease in the cost of building a minimal viable venture and further propelled by recent advancements in Generative AI. Founders currently need to stitch together multiple complex tools including payment platforms, support tools, tax calculation and remittance services, marketing and retention products, and more in order to support the regulatory, currency, and language requirements that come with selling in many geographies. A simpler option is starting to emerge.
The consumer personal finance market is home to dozens (if not hundreds) of startups that have helped their users, but haven’t been able to fully transform their financial lives. But now, thanks to generative AI, the much-discussed topic of “self-driving money” finally has a chance to achieve its potential.