Generative AI-powered shopping has been gaining steam in the latter half of this year and is shaping up to be one of the top trends as we move into 2026. The availability and convenience of Gen AI tools are shifting consumers’ shopping habits away from traditional browser-based shopping as these new tools become more deeply embedded in the shopping experience.

Adobe for Business published a report earlier this year that shows momentum in the use of generative AI-powered chat services. The report, which surveyed more than 5,000 US individuals, aimed to complement a study conducted during the 2024 holiday shopping season that examined consumers’ usage of Gen AI-powered chat services and browsers.

There are six standout findings from the report that exemplify the shift in consumers’ habits.

Generative AI traffic to retail sites exploded

Adobe found that the first surge during the 2024 holiday season reached a growth of 1,300% year-over-year. Following this, the recent data shows that traffic from Gen AI-powered chat tools and browsers has continued to accelerate, reaching a 4,700% year-over-year increase as of July 2025. AI-driven visits now represent a meaningful and fast-growing share of retail shopping and research activity.

Over one-third of US consumers have used AI for shopping

According to Adobe’s 5,000-person survey, 38% of consumers have used generative AI at some point during their shopping process, while 52% plan to do so this year. Shoppers are most commonly using it for product research (53%), recommendations (40%), finding deals (36%), creating shopping lists (30%), and gift ideas (30%).

AI shoppers are more engaged and informed

Consumers that arrive at a website from AI-powered sources spend 32% more time per visit, view 10% more pages, and have a 27% lower bounce rate than those coming from traditional sources such as search, social media, and email. Shoppers using Gen AI also report an increase in satisfaction, with 85% of users saying that AI improved their shopping experience and 73% now rely on it as their primary product research tool. Overall, shoppers using Gen AI tools are more engaged than shoppers using traditional ecommerce methods.

Conversions still lag

While consumers are increasingly using Gen AI tools for browsing, many stop there and fail to actually make the purchase. In fact, Adobe’s study showed that AI-driven traffic is still 23% less likely to convert than traditional traffic. This figure has actually shown a bit of improvement over the past few months. The study showed that conversion rates were 49% lower in January 2025 and 38% lower in April 2025. This suggests that consumers increasingly trust AI-powered recommendations enough to complete purchases directly.

Revenue-per-visit from AI sources is catching up to non-AI visits

When it comes down to dollar figures, it turns out that Gen AI-powered shopping isn’t as valuable, though that is quickly changing. Adobe’s study found that AI-driven revenue-per-visit rose 84% from January 2025 to July 2025. While AI-driven visits were worth 97% less than non-AI visits in July 2024, they were only 27% less than a non-AI visit a year later. This indicates shoppers are moving from using AI purely for research to actually buying through AI-driven paths.

Mobile is fueling AI-driven shopping growth

According to the study, in July 2025, 26% of AI-driven retail traffic came from mobile. This is up by 8 percentage points from 18% six months earlier. Adobe expects mobile AI use to further close the conversion gap, given consumers’ tendency toward more impulse-driven shopping on phones.

All of these statistics paint a picture of what we can expect to happen to ecommerce in the next few years. As consumers increasingly turn to their preferred Gen AI tool when they start their shopping journey, we’re witnessing the early stages of a new kind of marketplace. In ecommerce 2.0, we’ll see discovery, recommendation, and payment converge within a single interface. Competition in this new frontier will no longer be about who owns the checkout. Rather, it will centralize around who owns the conversation that leads there. As the 2025 holiday shopping season picks up, expect to see fintechs, retailers, and payment providers racing to claim their spot in the Gen AI shopping ecosystem.

---

Photo by Ivan Samkov

Stablecoins may have saturated headlines earlier this year, but September has marked a turning point to the industry. This month has brought four large announcements in agentic payments, demonstrating that the technology has moved from fringe to forefront.

And while the announcements speak volumes about how quickly technology developments move in fintech, it also sends seven major signals to banks and fintechs.

A preferred protocol layer emerges

Earlier this week, agentic commerce platform Circuit & Chisel landed $19.2 million to launch ATXP, a web-wide protocol. The protocol will not only position Circuit & Chisel as an orchestrator of agentic commerce, but it will also help streamline workflows and enable businesses to operate faster and more efficiently by leveraging revenue-generating autonomous agents.

The launch and growth of ATXP show the industry’s movement toward a web-wide standard for agentic payments. It also highlights how payments are shifting from app-specific functions into a common infrastructure layer.

Big Tech wants to lead

Google and PayPal made headlines last week when they announced their partnership on agentic shopping, embedded payments, payments processing, and more. The two are positioning themselves at the forefront of agentic payments and commerce and are providing developers with tools to engage in the new era of digital commerce.

The partnership between Google and PayPal shows that Big Tech wants to be at the forefront in shaping how commerce and payments flow online in the future. This early movement is a warning to players that sit back on the sidelines and wait for others to move first. Slow-moving banks and fintechs risk being relegated to backend providers unless they strategically find their own niche in the space.

Crypto and Web3 join forces with platforms

Also last week, Google announced that it is leveraging the x402 protocol within its Agent Payments Protocol (AP2) to allow AI agents to pay each other using stablecoins on Coinbase. With the ability to handle payments on behalf of their end users, agents will now be able to complete certain tasks that previously required manual oversight, such as paying for data crawls, services, or microtasks.

The launch merges crypto protocols and mainstream platforms, and is a great example of how agentic payments won’t be limited to decentralized finance environments. Instead, we’ll see agentic payments within web browsers, search, and commerce platforms.

Credit has an agentic future

After landing strategic backing from Citi Ventures earlier this month, agentic AI-powered credit data and payments platform Spinwheel plans to fuel growth, expand its agentic AI platform, build out its data sets and add new products. Additionally, Citi Ventures will advise the company on banking-specific product use cases.

This funding shows backing for the idea that consumer credit and agentic payments will be integrated in the future. It shows the breadth of potential for agents to manage payments, debt repayment, refinancing, and credit optimization.

The shift to autonomous decisioning

All four of these announcements demonstrate how payments will move from static, user-initiated tasks to autonomous, rule-driven events. To stay current, banks and fintechs will need to embed decisioning logic, risk scoring, and compliance into their payment flows.

Regulators will take notice

While regulators don’t have a lot of time (or expertise), agentic payments are sure to get their attention. These announcements around autonomous money movement have raised concerns around AML, KYC, and consumer protection issues. Firms that build compliance into agentic systems will be one step ahead in winning not only consumer trust but also regulators’ approval.

The race for standards is on

Much like open finance, the world of agentic payments will desperately need to abide by an agreed upon set of standards. Because competing protocols and ecosystems could fragment adoption, the disorganization could not only disrupt the user experience, but it could also wreak havoc on creating a clean, regulated environment. Whichever parties are involved in driving standards for payment rail interoperability will take the role that SWIFT did in shaping payments rails in the 1970s.

The ultimate question is, who will lead and who will follow?

---

Photo by Athena Sandrini

After the CFPB withdrew its lawsuit over Section 1033 of the Dodd-Frank Act, the bureau stated that it would begin a new, “accelerated” rulemaking process with an Advanced Notice of Proposed Rulemaking (ANPR) within three weeks. That three-week period ended last week, on August 22nd, when the CFPB published its Personal Financial Data Rights Reconsideration, effectively kicking off the new rulemaking process.

Much is riding on how this rule takes shape, not only for banks, but for fintechs and consumers alike. Visa’s recent move to abandon its US open banking initiatives underscores just how high the stakes are. In its latest release, the CFPB asked for comments and data to guide its decisions on four critical issues tied to Section 1033. Below, we’ll walk through each issue and explore the potential impact.

Representatives: who deserves access to the data?

The first of the four issues is defining who can serve as a representative on behalf of the consumer. The question essentially asks who can make a request to access the consumer’s data on their behalf. Today, this includes not only the consumer themselves, but also third-party aggregators and fintechs, as well. If the CFPB decides to narrow this scope, it could potentially block third-party services from accessing consumer data, limiting it to the consumer and the bank itself.

The latter would favor incumbents as it allows them ultimate control. For fintechs, this would create a risky environment. The uncertainty would make it risky to invest and build APIs that could be restricted in the future.

Fee structures: who pays for data access?

The second of the four issues seeks to determine the optimal amount of fees that banks should be able to charge in response to a customer-driven request. As a result, data access may no longer be free for aggregators, which may require them and fintechs to reshape their business models in response.

Charging for data would allow banks to recoup compliance costs for API access, but may receive negative attention from fintechs and consumers. Additionally, fintechs with already thin margins may be forced to look for an exit.

Data security: weighing threats vs. benefits

The third of the four issues the CFPB spotlighted is the threat and cost-benefit analysis for data security associated with complying with Section 1033. If the Bureau requires compliance with tighter security requirements, all stakeholders will feel the repercussions of tighter security expectations.

With tighter compliance, small fintechs that previously had limited compliance requirements may now need to step up to higher standards. This could ultimately lead to consolidation, since large, well-resourced firms would be able to more easily meet compliance.

Data privacy: the cost of protection

The final of the four issues the CFPB spotlighted is the threat landscape surrounding data privacy associated with Section 1033 compliance. The Bureau may set new limits on how fintechs are allowed to monetize consumer data in an effort to maintain their privacy.

With new guardrails on how they are allowed to monetize consumer data, fintechs may face limitations on using data for personalized marketing or other secondary data uses. As a result, innovation may slow down, but consumers may gain more confidence.

Your turn to comment!

The CFPB’s recent call for comments is more than regulatory housekeeping. It is highly consequential and will determine the future of open banking in the US. The Bureau’s questions signal real costs, risks, and opportunities.

It is important to make your voice heard on these issues! In the six days that the comment period opened, only seven comments have been submitted. Send your comments to the CFPB by October 10, 2025 at 11:59 pm EST.

---

Photo by Erik Mclean

Payments infrastructure company Stripe is moving into the blockchain, according to Forbes, which uncovered a job posting regarding the move. According to the posting, Stripe is planning to launch a payments blockchain called Tempo.

“Tempo is a high-performance, payments-focused blockchain,” the advertisement on the Blockchain Association’s website said. Here’s a look at five things that matter about Stripe’s move, including details about the new blockchain, why it’s launching it now, how it fits into the company’s strategy, what it means for the wider industry, and what’s still unknown.

What is Tempo?

Tempo is a Layer 1 blockchain built from the ground up (as opposed to a fork). A Layer 1 blockchain is the base network in a blockchain ecosystem. It serves as the foundational layer where transactions are processed, validated, and recorded. With Tempo, Stripe is optimizing the network for payments and making it compatible with Ethereum Solidity toolchains, meaning that developers can use the same set of familiar tools they use for Ethereum.

Tempo was built stealthily by a small team of around five people in partnership with crypto VC firm Paradigm. Until the job posting, which was dated August 3, the new project operated under the radar.

Why now? Stripe’s crypto build-out strategy

Launching its own blockchain is Stripe’s latest move into the crypto industry. Stripe has been steadily entering the crypto world, from its acquisition of stablecoin platform Bridge for $1.1 billion, to buying wallet developer Privy in June. Since then, Stripe has also made a non-crypto acquisition, acquiring payment orchestration company Orum in June. Launching Tempo will add the final piece of this equation. Owning its own blockchain rails will give Stripe full control of the payment flow, from the wallet to the payment settlement.

The benefits of building its own blockchain

As with all of its acquisitions, Stripe’s move to create a blockchain from scratch is strategic. Launching Tempo will offer it full-stack control, which will allow Stripe to optimize network speed, lower fees, and integrate with other stablecoins and wallets. Additionally, the custom payments blockchain could displace legacy systems like SWIFT or even FedNow, with faster, cheaper rails. And since Tempo will be compatible with Ethereum it is developer friendly, which means that it will not require new tools or talent to align with its infrastructure.

Bigger implications for payments & crypto

Stripe has been operating in the fintech arena since 2010. With its own blockchain, the company could accelerate mainstream adoption of stablecoins and blockchain payments via a merchant network. The move showcases how traditional fintechs are taking steps to operate in the crypto space. Not only this, but it is also indicative of a new competitive landscape in which fintechs control their own payments rails, disrupting traditional ecommerce and cross-border transactions.

What We Still Don’t Know

Even though it is interesting to speculate on the impacts Tempo will have across the industry, there is still a lot we do not know. Much of this is because the news originated from a job posting, not an official company announcement. Details such as whether Tempo will come with its own native token, how it will be governed, and a clear timeline for the launch are still unknown.

What is clear, however, is that it is worth keeping an eye on Stripe not just as a payments innovator, but also as a player in the crypto arena going forward.

The future of open banking in the US just hit another curve. Yesterday, July 29, the Consumer Financial Protection Bureau (CFPB) filed a surprise motion to pause the legal battle over its Section 1033 data access rule. The move, which comes the day that final arguments were due in the case, signals that the Bureau plans to rewrite the rule altogether, a dramatic shift that could reshape how financial data flows between banks, aggregators, and fintech apps.

Because this case is a rollercoaster, let’s start from the beginning. In the fall of 2024, the CFPB issued the final rulemaking on Section 1033 of the Dodd-Frank Act. The new rule, issued in the form of a 594-page document, aimed to enhance consumers’ rights, privacy, and security over their own personal financial data. Hours after the CFPB issued the rule, Forcht Bank, the Kentucky Bankers Association (KBA), and the Bank Policy Institute (BPI) filed suit, arguing that the CFPB overstepped its authority and violated the Administrative Procedure Act. They later filed an amended complaint in November, and the CFPB responded in December.

Then, things got even more complicated. After the new administration placed new leadership over the CFPB early this year, both sides agreed to temporarily pause the case and delay the rule’s compliance deadlines. The court granted the pause and extended it again in March. Meanwhile, the Financial Technology Association (FTA), a group that represents fintechs and aggregators, was granted permission to join the case in May. All sides submitted their main arguments to the court at the end of May, and their responses to each other’s arguments were due by July 29, 2025.

The next interesting twist in the story came out yesterday, when the CFPB filed for a motion to “reconsider the rule with the view to substantially revising it and providing a robust justification” because of “recent events.” The recent event the Bureau is referring to is JPMorgan’s announcement to data aggregators that it plans to charge them a fee to access consumer data. This move has sparked multiple heated conversations over who owns consumer data and how many services banks should be expected to provide for free.

Many conversations highlighted the fact that the CFPB’s 1033 ruling stipulated that banks cannot charge third parties for data access. However, now that the CFPB has indicated it plans to revise the rule, some experts are wondering whether the agency is backing away from that stance. Critics fear the Bureau may soften the ban on data access fees in response to pressure from JPMorgan, the biggest bank in the world by market capitalization. Others argue that resetting the rule could allow for broader industry consensus and a more durable framework.

Either way, the move reopens a regulatory debate that many thought was settled. Now that you’re caught up, here are the four highlights of the CFPB’s latest motion:

1. Plans to rewrite the rule
   In its motion, the CFPB announced that it intends to throw out its current version of the Section 1033 rule and start fresh. The Bureau says it will kick off a new, “accelerated” rulemaking process with an Advanced Notice of Proposed Rulemaking (ANPR) expected within three weeks. The CFPB has decided on a pivot. Rather than defending its original rule in court, the CFPB is acknowledging its change in leadership and “recent events in the marketplace” (which is very likely JPMorgan’s controversial move to charge data aggregators) as the reason it plans to take the rule back to the drawing board. A substantially revised rule could reshape the boundaries around consumer data access and fees.
2. Requests a pause
   The CFPB is asking the court to pause the current legal proceedings while it works on rewriting the rule. The Bureau reasons that if it is about to replace the rule that’s being challenged, then it doesn’t make sense for the court to keep spending time on the existing version.
3. Promises communication
   To assure the court that the motion isn’t a stall tactic, the CFPB is promising transparency throughout the rulemaking process. If the court grants the pause, the Bureau will submit status reports every 90 days to update the judge on its progress. This is meant to maintain communication with the court and demonstrate that the agency is moving swiftly.
4. Cites opposition
   In the motion, the CFPB acknowledges that not all parties are on board. The Financial Technology Association (FTA), which represents data aggregators and fintech firms, does not oppose the pause. But the plaintiffs (Forcht Bank, the KBA, and the BPI), however, do oppose it. They plan to file a formal objection, likely arguing that the pause is a delay tactic and that the rule should be struck down based on the current legal merits.

What happens next will most certainly have ripple effects across the entire financial ecosystem. If the CFPB follows through on its promise to rewrite the rule, we could see a very different version of Section 1033, one potentially shaped by politics, institutional pressure, consumer rights, and innovation. At this point, it is clear that the future of open banking in the US is once again uncertain and very much up for debate.

---

Photo by Christian Wasserfallen

With the first half of 2025 behind us, it’s a good time to look forward to what the second half of the year will bring. The first two quarters were packed with change: from the stablecoin frenzy and cuts to the CFPB in the US, to new regulatory crackdowns across Europe and the reversal of Section 1033, reshaping the future of open banking. Meanwhile, banks and fintechs are ramping up their use of AI, navigating new regulatory requirements, and adapting to global momentum around real-time payments and digital identity.

With all of this change, it’s hard to imagine the surprises that the next two quarters will bring. And while I can’t predict all of the surprises, there are five trends that banks and fintechs should not ignore as we move into the second half of the year.

The open banking conversation evolves

In the EU, PSD3 and the Financial Data Access (FIDA) framework are being finalized and the UK is moving forward with Open Banking 2.0 under the Joint Regulatory Oversight Committee (JROC). In contrast, the US is in a period of regulatory uncertainty. The CFPB is pulling back from Section 1033 and JPMorgan revealed to data aggregators that it plans to increase the cost for them to pull consumer data. Banks need to keep a close eye on the evolving conversations around open banking as ripple effects take place across the globe.

AI becomes an arms race in financial services

AI is quickly becoming table stakes for financial services organizations. AI-native fintechs are setting new expectations around service, automation, and personalization. And firms are no longer stopping at chatbots and GenAI technologies. Instead, banks across Europe, the US, and Asia are increasingly integrating agentic AI, and even hiring AI agents for tasks like underwriting, compliance, and customer service. Expect the second half of the year to bring a continued rise in AI literacy programs and internal tooling as firms upskill teams and reduce reliance on third-party vendors by turning instead to agentic AI.

Tokenization takes over

In the first half of 2025, we saw major pilots for tokenized deposits, treasuries, and real-world assets (RWAs). In the latter half of the year, we can expect to see real world implementations, particularly in wholesale payments, interbank settlement, and liquidity management. Regulatory clarity is also beginning to transpire. Jurisdictions like the EU, Hong Kong, and Singapore are starting to define legal frameworks for tokenized financial products. This may prompt US regulators to clarify the treatment of tokenized deposits and securities.

Identity verification becomes a battleground

With rising fraud, easy-to-create deepfakes, and an increase in embedded finance, financial institutions are shifting from one-time identity checks to continuous, context-aware identity verification. The second half of this year will bring increased adoption of reusable digital IDs, decentralized identity frameworks (DID), and advanced biometrics tied to behavioral signals. As always, the challenge will be balancing a low-friction user experience with high security.

Real-time payments reshape expectations

FedNow is gaining traction in the US, ISO 20022 began rolling out earlier this week, and stablecoin-powered cross-border projects are on the rise. All of these aspects, plus an increase in stablecoin adoption are making real-time payments the norm and are raising customer expectations. Banks that can’t meet those expectations risk losing ground to more nimble players.

---

Photo by Pixabay

At this point, if you’ve been working in the financial services industry since January, you’ve likely heard of stablecoins, and you may have heard of tokenized deposits. What may still be unclear, however, are the differences and similarities between the two.

Blockchain-powered financial infrastructure is on the rise, and it’s important for banks, fintechs, and regulators to understand new developments in the space, what’s possible, and what’s next. Here’s a brief overview of where stablecoins and tokenized deposits intersect, where they are different, and where they may be most useful.

Stablecoins

Stablecoins are digital assets that are issued by private companies or protocols and pegged to fiat currency. Some of you may be familiar with are Circle’s USDC, Tether’s USDT, and PayPal’s PYUSD. It is important to note that stablecoins are backed one-to-one by off-balance-sheet returns, such as fiat cash or Treasuries. Unlike fiat held at a traditional financial institution, however, they are not FDIC-insured.

Tokenized deposits

In contrast, tokenized deposits are bank-issued digital representations of fiat deposits, recorded on a blockchain. The deposits sit on the bank’s balance sheet, are fully integrated into the bank’s infrastructure, and are minted and backed by regulated banks.

Differences

There are key differences between stablecoins and tokenized deposits. First, let’s look at the issuer. While not always the case, most stablecoins are issued by private, non-bank companies. Even though some banks have issued “coins,” as in the case of JPMorgan’s JPM Coin, they are considered tokenized deposits and are usually used internally for payment settlement, not open to the public, and are not tradable on public blockchains.

The backing structure of stablecoins and tokenized deposits is also different. For example, stablecoins are not held on the bank’s balance sheet and represent a one-to-one reserve of fiat currency. In contrast, tokenized deposits are held on a bank’s balance sheet. This is useful when a firm wants to maintain liquidity to support lending and credit creation, and ensure that customer funds are protected in a regulated financial institution.

Speaking of regulation, FDIC insurance is a key differentiator between stablecoins and tokenized deposits. Stablecoins currently operate in a developing regulatory environment and, importantly, they do not offer deposit insurance such as FDIC. Tokenized deposits, on the other hand, are both insured by the FDIC and regulated.

Another key differentiating factor between the two blockchain-based payment tools is that they have opposite effects on liquidity. Stablecoins remove liquidity. That’s because when consumers exchange their fiat currency in exchange for stablecoins, their fiat currency leaves their wallet and sits in reserves, generally in the form of safe, passive assets like US Treasuries or custodial accounts. This reduces the money multiplier effect and may even weaken bank balance sheets over time. In contrast, tokenized deposits stay on the bank’s balance sheet, making the funds usable for lending, investing, and general liquidity management.

Use cases also differ between stablecoins and tokenized deposits. While stablecoins are best known for their use in cross-border payments, programmable payments, and in DeFi. Tokenized deposits are useful for domestic real-time payments, B2B payments, and treasury automation.

Similarities

But though they differ in all of these aspects, there are also a handful of similarities between stablecoins and tokenized deposits. First, both are programmable, blockchain-based representations of fiat currency. However, it is important to distinguish that, while stablecoins are backed by dollars (fiat currency), tokenized deposits are actual, digital representations of dollars.

Next, both can be used to enable payments and reduce settlement times. Because they take place on the blockchain, transactions in both stablecoins and tokenized deposits can take place in near-real-time. This eliminates the delays associated with traditional clearing and settlement systems, which can take up to three business days. Whether it’s a purchase, B2B payment, or interbank transfer, blockchain-based transactions allow for faster value exchange.

Additionally, both can be used in smart contracts, programmable payments, and embedded finance applications. And while tokenized deposits aren’t commonly used in the DeFi economy at the moment, that may change once regulated or institutional DeFi networks become more common.

Finally, stablecoins and tokenized deposits alike are useful for modernizing payment rails. Already in their infancy, both are acting as gateways to more advanced financial infrastructure. By enabling real-time, programmable payments on blockchain networks, they help move the financial system away from slow, batch-based legacy systems like ACH or SWIFT.

The future of both

Looking ahead, it is possible that stablecoins and tokenized deposits will coexist, as they both serve different niches. No matter which structure reigns supreme, however, we will certainly see traditional financial institutions and private DeFi companies increase their focus on interoperability and shared infrastructure. As regulatory clarity is enhanced on both sides and new pitfalls are discovered, the industry will likely converge on a hybrid model that blends the safety of traditional finance with the speed, transparency, and programmability of decentralized infrastructure.

The GENIUS Act passed in the US Senate yesterday with a 68 to 30 vote. The bill now moves to the House, where it’s up against the STABLE Act. This means that the House will need to choose between passing the GENIUS Act at face value or passing and reconciling the STABLE Act. 

For financial services, the GENIUS Act is a big deal. That’s because it is not only the first stablecoin legislation to gain real bipartisan traction, but it will also serve as a foundation for the US to begin a digital asset ecosystem. Overall, there are four major implications the bill has on banks.

Stablecoins gain legitimacy and clarity

As a decentralized finance tool, stablecoins have long been grouped together with their crypto cousin bitcoin. Because of this, many traditional financial institutions in the US have shied away from associating themselves with stablecoins.

The GENIUS Act, however, offers both banks and fintechs a clearer legal framework to issue and use stablecoins since it outlines requirements for licensing, reserves, and oversight. Having regulation on their side reduces regulatory uncertainty and will encourage financial institutions to adopt the new payments tool and leverage stablecoins for new use cases. Reducing ambiguity around compliance and risk will also benefit firms exploring tokenization.

Banks may face new competition from Special Purpose Depository Institutions

The Senate version of the bill includes a controversial provision allowing Special Purpose Depository Institutions (SPDIs), such as Kraken, to operate across US states without the approval of each host state’s banking regulator.

If the bill is successful, it will allow fintechs with SPDI licenses to gain a regulatory shortcut because they do not need to comply with capital and liquidity requirements. This may erode the role of traditional banks in certain payment and custody markets and may not be a positive change.

“That is a pretty significant expansion of special purpose depository institutions,” Klaros Group Partner Michele Alt told American Banker. “I would ask, what else could you create as a special depository institution? How could this be used?” 

Notably, however, even though the bill has passed through the Senate, the House’s version of the stablecoin bill doesn’t include a similar provision. This means that if the bill does pass through the House, the House and the Senate will need to convene for a conference to come to an agreement. 

Rising expectations for real-time money movement

While consumers already expect many things in real-time, the GENIUS Act adds more pressure for banks and fintechs to deliver faster, more programmable payments. The bill will enable regulated stablecoins and essentially facilitate real-time settlement, 24/7 money movement, and programmable financial interactions.

This method of funds transfer won’t rely on traditional rails like ACH, wires, or even FedNow. If end users and businesses get accustomed to real-time, programmable payments, their expectations may be permanently shifted, requiring banks to keep up.

This adjustment would be tricky for banks, as many would need to invest in infrastructure that supports tokenized payments, smart contracts, and on-chain compliance.

Banks need to stay agile

If the House does not pass the GENIUS Act, it can advance its own bill in the form of the STABLE Act or negotiate a compromise. Either way, regulatory change is clearly in motion. Banks and fintechs should closely monitor the developments and begin scenario planning now. Whether it’s the GENIUS Act, the STABLE Act, or a hybrid outcome, stablecoin regulation is on the horizon. Those who prepare early will be best positioned to compete in a tokenized financial future.

---

Photo by Andrew George on Unsplash

Agentic AI agents, autonomous agents that act on behalf of users with minimal input, are not just coming to financial services. They’re already here. One of the most compelling use cases for Agentic AI is at checkout, where commerce, AI, and payments converge at the point where consumers make their purchase decisions.

In the past few weeks, three Agentic AI shopping and checkout announcements from major payments and technology players have made news headlines. So far, Google, Visa, and Mastercard are leading the Agentic AI payments charge, with PayPal and Perplexity not far behind. Here’s a look at what each company is doing.

Google’s AI-powered shopping agents

Google announced its AI Shopping Mode yesterday, a new online shopping experience that allows users to browse 50 billion product listings and buy the item they want using Google’s new agentic checkout at a price that fits their budget. Shoppers set their preferences by selecting “track price” on a preferred product listing and set the right size, color, and the amount they want to spend. If the item’s price drops into the user’s pre-selected price range, they receive a push notification and can have the agentic shopping agent buy the item for them with the push of a button.

Google is embedding an AI assistant into every step of the purchasing process, from browsing to payment, and is making the checkout experience hyper-personal, with less friction.

Visa’s intelligent commerce and agentic AI

Visa unveiled its Visa Intelligent Commerce tool last month. The new initiative will empower AI agents to deliver personalized and secure shopping experiences for consumers at scale. The program will equip AI agents to seamlessly manage key phases of the shopping journey, from product discovery, to purchasing, to post-purchase product management.

Unlike Google, Visa will offer APIs and SDKs that will provide third parties a suite of payments tools, including tokenization, authentication, and transaction controls, to embed into their own apps. In this sense, Visa is not just planning to launch a new checkout tool, it is building infrastructure for a world where the AI agent is the end customer.

Mastercard’s agentic payments through Agent Pay

Mastercard announced Agent Pay, a payment framework for agent-driven commerce, 24 hours before Visa’s agentic AI announcement hit the wires. Mastercard’s tool aims to make payments smarter, more secure, and more personal by embedding them directly into the product recommendations generated by GenAI platforms.

When paired with Mastercard’s tokenization technology, Agent Pay will not only add security, but will also help retailers identify and validate customers to offer a more meaningful and consistent shopping experience. Overall, Mastercard is pioneering a payment model where AI, not the consumer, initiates the purchase.

Perplexity x PayPal

Earlier this month, GenAI-powered search engine Perplexity partnered with PayPal to enable in-chat shopping. Shoppers will be able to check out instantly with PayPal or Venmo when they ask Perplexity to find a product, book travel, or buy tickets. The entire process will be powered by PayPal’s account linking, secure tokenized wallet, and emerging passkey checkout flows, which could eliminate the need for passwords.

While it is not a formal “agentic” platform, the move shows that large language models (LLMs) are starting to transact directly and the partnership is a good example of how chat interfaces are evolving into commerce platforms. The announcement serves as a preview of agentic commerce where LLMs initiate and complete purchases in a single conversational flow.

Overall, these announcements signal a major shift in ecommerce. The online point-of-sale is moving from a consumer-initiated process to an AI-initiated transaction. At the outset, regulation, identity, fraud, and explainability will be a large challenge. Still, the shift to agentic commerce is well underway, and the companies building today’s infrastructure are setting the rules and structure for how agentic AI commerce will work in the future.