Author: IreneDu
This is part 2.5 of the Stripe AI Strategy Deep Dive series.
The series originated from observing the 288 products announced at Stripe Sessions 2026 on April 30. I observed that Stripe is attempting to become the economic infrastructure for the AI Agent era.
The first piece, Stripe Is Not a Payments Company, attempted to answer "Why Stripe"—its DNA determines it can do this.
The second piece, KYC Is Dead, the Agent Economy Is Rewriting the Underpinnings of Financial Regulation, wanted to dissect the future Stripe is betting on—what exactly the Agent economy looks like, and why traditional payments infrastructure will completely fail in front of it.
But while writing the second piece, I received a comment from a peer:
I completely agree with the first part. AB 316, or any sovereign nation's laws, will not acknowledge in the short term that an "Agent is a legal entity"—the ultimate defendant will always be a specific person. This is something Know Your Agent cannot change, and it shouldn't try.
But for the latter part—"The only change is payment and settlement efficiency"—I have reservations. The issue with this statement is not the conclusion, but the default framework it assumes: it sees KYA as an upgrade to the existing payments system.
This is precisely what I believe is worth writing an extra piece to discuss.
First, let's return to the muscle memory of a former payments practitioner:
Payment forms are scenario-driven; they are not designed from within the payments system.
Every true leap in payments—online banking, mobile wallets, QR code scanning—was not because someone built a better product within the payment layer, but because a new transaction scenario emerged that shattered the underlying assumptions of the previous payment system.
New payment forms "grow" from the infrastructure required by that scenario; they are not "optimized" into existence.
I once worked on payments innovation at Ant. On a platform that was the absolute industry pioneer in creating "Quick Pay," "Mobile Pay," and "QR Code Pay," the greatest pleasure and pain was pondering: What is the next generation of payments?
We worked on watch payments (and heartbeat verification to replace facial scanning), NFC payments (the primitive technology behind "tap-to-pay"), participated in and authored several "next-generation" payment protocols, and even tried to get leadership to support my exploration of metaverse payments.
Most of these projects didn't take off.
Looking back, the reason was the same: we tried to define new payments within the payment layer, but the scenario driving the payments revolution hadn't arrived yet—without the scenario, the infrastructure the scenario needs cannot grow, and no amount of clever design in the payment layer can catch it.
The Agent economy is that new scenario I was desperately waiting for in the past.
KYA is the layer of infrastructure that is now growing.
KYA is not a product in the payment layer; it is the infrastructure layer for the Agent economy.
The five layers of KYA I defined in the previous article—Agent Identity, Authorization Scope, Intent Signing, Liability Chain Auditing, and Credit Rating—only Authorization Scope and Liability Chain Auditing fall on the payment chain. The other three layers (Identity, Intent, Credit) are not in payments at all.
- The Identity layer serves all scenarios requiring Agent identification: cross-platform calls, regulatory filings, internal corporate audits—payments is just one of them.
- The Intent layer serves the larger issue of AI alignment—payments is just one of its many verification scenarios.
- The Credit layer serves any system needing to assign permissions and quotas to Agents—payments is again just one of its users.
Therefore, that peer's judgment—"The only change is payment and settlement efficiency"—translated into the language of infrastructure means: viewing KYA as a subsystem of payments.
My judgment is the opposite: payments are a subsystem of KYA.
This reversal is the core of this piece.
The investment actions of Stripe, this company at the industry frontline, happen to be the empirical evidence.
The term Patrick Collison used at Sessions 2026 wasn't "AI payments," it was "economic infrastructure for AI." This isn't marketing speak; it's a positioning choice. It shows Stripe doesn't intend to lock itself into the identity of a "payments company"; it's betting on building the foundation for the Agent economy.
Specifically in product layout:
The Agentic Commerce Protocol (ACP) co-built by Stripe and OpenAI, now used by Microsoft Copilot, Meta, and Google Gemini which joined in April—is essentially an identity and session protocol, not a payment protocol.
Shared Payment Token, which separates the Agent from the real card number, operates at the authorization layer, not the settlement layer.
Stripe acquiring Bridge for stablecoin infrastructure, acquiring Privy for embedded wallet capabilities, and building its own Tempo blockchain for settlement pipelines—this entire layout doesn't fit within the framework of "payment efficiency optimization."
This kind of investment portfolio only makes sense under the judgment that "KYA is the infrastructure layer." If the Agent economy were merely a payment efficiency problem, Stripe wouldn't need to do stablecoins, embedded wallets, or build its own L1. What it's doing is securing positions layer by layer within that five-layer KYA.
Several numbers given by Stripe's Data Lead, Emily Glassberg Sands, in an Every interview in April this year, confirm the same thing from another angle: a large AI client is blocking 250,000 fraudulent free trials per week; she saw an AI company burning $25 in compute per free trial with a 4% conversion rate, meaning losing $625 to acquire one paying user; overall abuse of free trials has quadrupled in the past six months.
These numbers collectively illustrate one thing: in the AI economy, the judgment that truly determines whether a transaction can succeed and is worth doing no longer happens at the moment of checkout—it happens further upstream, with questions like "Who is this?", "What do they want to do?", "Is this worth allocating resources to?". This is why Stripe is moving its risk control Radar from the "transaction moment" to the "entire user lifecycle": it's not about making old risk control faster, but changing the questions risk control cares about from "Is there a problem with this payment?" to "Is there a problem with this user's/Agent's entire behavior?". The former is a payment layer issue; the latter belongs to the KYA layer.
Returning to that peer's question: where does liability ultimately fall?
He is correct—the ultimate legal entity remains a person. AB 316 has already codified this legally.
But this is precisely the real problem KYA must solve: when the liability chain becomes distributed, finding "specifically which part on which person" is something the KYC era didn't need to do, but the KYA era must.
In the KYC era, the liability chain was linear (user → payment/bank → merchant). When a transaction had a problem, you intuitively knew who to go to.
In the KYA era, the liability chain is a network (user → Agent platform → model provider → payment protocol → bank → merchant, potentially calling other Agents in between). Even if the law tells you "go after the person, not the Agent," you still don't know which person—because liability is already distributed across 5–7 entities.
KYA cannot change the law's ultimate assignment. But it can, within the network chain, use cryptography to solidify the role and actions of each entity—who authorized what, who executed what, who settled what, who fulfilled what. Turning "can't find evidence" into "can find evidence"; turning "which link had a problem is unverifiable" into "verifiable."
This is not an improvement in payment efficiency.
This is the first time accountability traceability can occur within an Agent network.
Therefore, I believe the statement "The only change is payment and settlement efficiency" confuses infrastructure with function.
What's really happening is:
- Because a new type of economic actor (Agent) has emerged, a new layer of infrastructure (KYA) is forced to grow;
- This infrastructure layer redefines "who is on the other side, what can they do, and who do we go to if something goes wrong"; on top of this infrastructure layer, payments will reorganize themselves in a form we cannot fully see today.
What exactly is the next generation of payments? What remains unclear is precisely the new species Stripe is attempting to define.
But in a world of uncertainty, one thing I am certain of—it will not be designed within the payment layer.
It will grow from the scenarios once the KYA infrastructure layer is laid.
