Payment networks have spent the past year racing separately to make AI agents capable of paying for things. This week, the biggest names in that race agreed to slow down and build the rules together first.
A coalition, not a product launch
The Agentic Payments Alliance (APA), announced August 18 by stablecoin infrastructure firm Rain, launches with more than 25 founding members spanning the parts of finance an autonomous agent would need to touch to transact: card networks (Visa, Mastercard), payment processors (Fiserv, Evertec, Shift4, Lithic), stablecoin and blockchain infrastructure (Circle, Solana, Avalanche, Monad, Uniswap Labs, Fireblocks, Chainalysis), and fraud and identity specialists (Sardine, Turnkey, Episode Six). It is structured deliberately as a working coalition rather than a company or a standards body owned by any single founder: members set the charter and mission collectively, and none of them, including Rain, controls the outcome unilaterally.
That structure is the story. Agentic commerce, AI systems initiating and completing purchases on a person’s behalf, has moved from pilot programs to real transaction volume faster than the infrastructure underneath it has matured. McKinsey projects the category will handle between $3 trillion and $5 trillion in global commerce by 2030. Until this week, the companies building the rails for that volume, authorization, fraud liability, loyalty portability, were each solving those problems independently, which is exactly the setup that produces incompatible standards a few years in.
Advertisement
What the alliance is actually building
The APA’s initial mandate covers three concrete workstreams: shared research and technical frameworks for how an agent proves it is authorized to spend on a user’s behalf, testing standards for agent identity, and coordinated regulatory advocacy as lawmakers start asking who is liable when an autonomous agent makes an unauthorized purchase. It has also launched an Agentic Startup Program, giving five early-stage companies structured access to the alliance’s members and frameworks rather than requiring them to negotiate integrations one card network and one processor at a time.
Mastercard’s involvement is a useful signal of how seriously the network operators are taking the coordination problem. Sherri Haymond, the company’s executive vice president and global head of Digital Commercialization, tied the move to Mastercard’s long-standing role in setting shared commerce standards: “For decades, Mastercard has helped shape the standards that enable commerce at scale, and our participation in the Agentic Payments Alliance is a natural extension of that work for the agentic era.” That framing matters because Mastercard and Visa are direct competitors on nearly everything else in payments; their shared presence on a single standards table signals both companies have concluded that fragmented agent-authorization protocols would be worse for the entire category than the cost of the two of them agreeing to common ground.
The gap the alliance is racing to close
Rain’s own framing of the APA’s founding was explicit about the risk of moving too fast without alignment. Farooq Malik, Rain’s co-founder and CEO, said the alliance exists because “no single company should get to decide how agents transact on someone’s behalf. That has to come from the platforms building the rails, the regulators setting the rules, and the innovators closest to how agents are actually being used today.” He added that the group was formed “to put all of these parties in the same room, and to do it now, while the category is still taking shape.”
The parts of the stack still genuinely unsettled are the ones with the most liability exposure: how an agent proves it has standing authorization from a human, how a merchant or issuer catches a compromised agent before it drains an account, and how loyalty points or rewards attach to a purchase an AI system initiated rather than a person clicking checkout. Those are the same categories of infrastructure risk this publication has tracked as banks build tokenized settlement rails, including the shared ledger work Bank of England has been running with participants like Polygon Labs on its digital pound trial, where the hard problem has likewise been getting competing infrastructure providers to agree on shared rules before scale forces a messier standard on everyone.
What it means for the payments leader
For a bank or processor evaluating whether to support agentic checkout, the APA’s existence is a signal to wait for its output before building bespoke authorization logic in-house. A proprietary approach to agent authorization risks becoming incompatible with whatever the alliance’s members converge on, the same trap that has slowed stablecoin settlement infrastructure until enough of the market agreed on shared rails. Fintechs building agent-facing commerce tools should treat APA membership, or at least alignment with its published frameworks once they land, as a proxy for which authorization and fraud-detection standards are likely to become de facto requirements for merchant acceptance.
The alliance has not yet published technical specifications, only its founding mandate and workstreams, so the near-term test is whether more than 25 companies with competing commercial interests can actually converge on shared standards before regulators or a dominant platform impose one for them. Mastercard and Visa sitting at the same table is evidence they think that convergence is achievable. Whether it happens before the first major agentic-commerce fraud incident forces the industry’s hand is the open question the APA now exists to answer.
History offers a reasonable template for how this plays out. Card networks have coordinated on shared standards before, most visibly through EMV chip specifications and later through tokenization frameworks for mobile wallets, both cases where competitors concluded that a common technical baseline grew the overall pie faster than any one company’s proprietary approach could. Agentic commerce differs in one important respect: the liability question is not yet settled anywhere in the world, which means the APA’s regulatory advocacy workstream may end up mattering more than its technical one. A shared authorization standard is only useful if regulators agree on who eats the loss when it fails, and that conversation is still largely unresolved on both sides of the Atlantic.
For now, the practical signal for the industry is that the APA’s founding roster reads like a checklist of the vendors most fintechs already integrate with for card processing, stablecoin settlement, and fraud screening. Any framework the alliance eventually publishes will arrive with built-in distribution across that member base, which is precisely why waiting for it, rather than building a proprietary agent-authorization layer in parallel, is likely to be the more durable strategy.
Source: PR Newswire
