They're Already on the Next Layer

The GENIUS Act rulemaking deadline is July 18. That is seventeen days from today.

Four tokenized deposit networks have launched, committed to launch, or entered pilot in the last ninety days. SoFi Bank launched SoFiUSD in May (the first stablecoin issued by an OCC-chartered national bank on a public blockchain, available to its fifteen million members). Major bank consortia are committed to programmable deposit infrastructure. Regulatory comment periods are closing.

The next layer of money movement is not a concept anymore. It is infrastructure that is happening whether your institution is ready or not.

Many financial institutions are watching this closely. But not many are asking the question that matters most: do we have a foundational instant payments strategy.

Tokenized deposits and regulated stablecoins run on distributed ledger infrastructure, not on FedNow or RTP. (That is an important distinction, and I’ll come back to this later.) But the fiat that backs them, the reserves that settle redemptions, the deposits that fund the whole system, those still flow through the traditional payment rails your financial institutions use today. FedNow. RTP. The ACH network. Correspondent relationships that connect domestic infrastructure to the international networks like SWIFT.

So these worlds are not separate. When a stablecoin is redeemed for dollars, the fiat moves on traditional settlement infrastructure. When reserves are replenished, deposits clear through your core. The programmability lives on the blockchain layer. The fiat settlement lives beneath it. Both layers need governance, and they are not independent of each other.

Programmable money is money that executes on conditions. A tokenized deposit settles when a verified event occurs. A stablecoin-backed payment moves when an instruction is authenticated and authorized. An AI agent initiates a treasury disbursement when portfolio conditions trigger a rule. At the end of the day. In every case, when the fiat reaches a bank account, it still moves through traditional payment infrastructure. The programmability sits above the rail. The governance that makes it safe sits at the rail.

The governance gap that instant payments exposed, the need for real-time, pre-settlement decisioning, is the same gap that programmable money in all its forms will widen (this is because they all share the same underlying problem: money moves before humans review it, and your controls have to work in that environment regardless of which layer you’ve set the rules).

When financial institutions add instant payment send capability, they have to address the inherent decisioning challenge that didn’t exist in a batch environment. Before money can move irrevocably, the financial institution has to make a trust determination in real time. Is the sender authorized? Is the receiver verified? Is this transaction consistent with established patterns? Does the liquidity position support it? Those questions have to be answered before the payment clears, not after.

That real-time decisioning architecture, the control layer between the core banking system and the payment rail, is what governs instant payments. It is also what will govern the fiat flows that tokenized deposits and regulated stablecoins depend on when they operate at scale.

It’s important to be precise here, because the conflation of these systems is a real credibility risk. Tokenized deposits live on the bank’s own distributed ledger. Stablecoins like SoFiUSD live on public blockchains. Neither moves on FedNow or RTP. But both are anchored to the fiat system. Redemptions settle through banking infrastructure. Reserves are held in bank accounts. The governance layer you build for instant payments is the same governance layer that fiat-backed programmable money will rely on.

This is not a new build. It is an extension of existing infrastructure. But it can only be extended if it already exists.

The financial institutions racing to participate in the next layer are making an implicit assumption: that the governance foundation beneath them is ready to hold more weight. For some, that assumption is correct. For many, it is not yet true.

Here is a scenario that could already be happening.

A commercial client’s treasury platform just upgraded. The AI agent managing their cash positioning now sweeps overnight balances into a tokenized deposit product. Still fiat-backed. Still looks like a normal treasury move. But the instruction is AI-authorized, executing on a condition, not a human approval. And it settles instantly and irrevocably on your rail.

Your institution did not build anything. The programmable money touched you anyway.

This matters because when a human approved that sweep, your fraud controls had something to work with: behavioral biometrics, device fingerprint, session context, hesitation markers, the pattern of how a person navigates before they confirm a transaction. An agent has none of that. It executes the moment a condition is met, cleanly and consistently, which looks identical whether the instruction is legitimate or compromised.

The signals your controls are looking for are not there. The signals that actually matter now are different ones: is this agent registered and known to your institution, is it currently authorized by the client it is acting for, is it operating within the scope it was given. Most institutions are not collecting those yet.

And there is no safety net beneath it. Token rails have no return code, no chargeback, no recall window. The feedback loop that trained your fraud models on chargebacks and ACH returns does not exist here. You do not just lose the human signals. You lose the answer key.

That collapses everything into one moment. The real-time decision before the payment clears is no longer your first line of defense against fraud. It is your only one. And most institutions are still running detection logic built for a human-speed, human-initiated world.

The platforms building the network infrastructure do not solve for this. It sits with your institution.

The GENIUS Act rulemaking deadline has been framed primarily as a regulatory compliance milestone. Banks and nonbanks preparing to issue payment stablecoins need to understand the rules. That is only part of the story. (Note: agencies have a statutory obligation to finalize rules by July 18, but they face no formal penalties for missing the deadline. The GENIUS Act takes effect on the earlier of 18 months from enactment, which is January 2027, or 120 days after final rules drop.)

July 18 brings a deeper architectural question: Can your institution operate in an environment where programmable money, stablecoin-backed, tokenized, AI-authorized, interacts with your controls in real time?

A few leadership questions worth asking before that date:

• Does your fraud posture currently extend to your instant payment send capability, or does it operate reactively on the back end of a transaction that has already settled? If a programmable instruction executes automatically and irrevocably, where in your control architecture does the risk decision occur?

• Does your liquidity visibility operate continuously, or does it still depend on end-of-day batch reporting? Programmable money that executes on conditions does not pause for settlement windows. If your treasury view runs on batch data, you are managing a 24/7 rail with a 9-to-5 instrument.

• Does your institution have a framework for AI-authorized transactions? Automated treasury disbursements are already operating inside commercial banking relationships. When an AI agent initiates an instant payment on behalf of a commercial client, what governs that interaction on your side of the transaction?

These are questions that need to be answered, whether or not your institution is the one launching the programmable money products.

The institutions best positioned for the programmable money era are the ones that governed the underlying rail.

When the control layer is in place, the next layer of capability becomes an extension, not a reconstruction. Tokenized deposits can operate within existing fraud and identity controls. Regulated stablecoins can route their fiat settlement through existing multi-rail governance. AI-authorized transactions can be evaluated against existing risk parameters. The new products sit on top of the infrastructure. The infrastructure holds.

The financial institutions that took governance seriously before they scaled instant send are positioned to build the next layer without rebuilding from scratch.

That is the compounding advantage of sequencing correctly.

The programmable money moment is here. The question is not whether to participate. It is whether what you built for instant payments can carry what comes next.

Here is the question worth taking into your next board conversation: if the next layer launched tomorrow, would your governance architecture hold the weight?

If your leadership team would benefit from this perspective in the room, find me on LinkedIn or at fintech-consultant.com.

Share The Instant Edge

© 2026 FinTech Consulting LLC. All rights reserved.

The Instant Edge is published by Marcia Klingensmith, Payments Maven™.

Payments Maven™ · The Instant Edge · July 1, 2026 · Instant-Enabled Banking