On 18 August 2026, Rain [launched the Agentic Payments Alliance](https://www.rain.xyz/resources/rain-launches-the-agentic-payments-alliance) with 26 founding members, Visa and Mastercard among them. It's a working coalition rather than a company: members set the charter together, and the early agenda is shared research, frameworks for agent identity and authorization, and advocacy on the regulatory questions agent commerce keeps raising. It can settle how two systems talk to each other about an agent's purchase. It can't settle what your agent is allowed to buy.
Most of the coverage stopped at the logos. Start instead with who signed, because the composition of the list says more than the press release does.
Who is actually in it?
All 26, in full: Avalanche, Basis Theory, Chainalysis, Circle, Coinflow, Crossmint, delta Network, Episode Six, Evertec, Fireblocks, Fiserv, Kala, Lithic, Mastercard, Monad, PayOS, Rain, Remitly, Rialo by Subzero Labs, Sardine, Shift4, Solana, Turnkey, Uniswap Labs, Visa and Yuno.
Sort that by rail and it splits almost in half. Fiserv, Shift4, Evertec, Episode Six and Lithic are card infrastructure. Circle, Solana, Avalanche, Monad, Fireblocks, Turnkey and Uniswap Labs are crypto infrastructure. Chainalysis and Sardine sit on risk. Remitly moves money across borders.
Rain, the convenor, issues cards against stablecoin balances, which is roughly the seam the alliance is built along.
Both card networks sitting down with the stablecoin issuers is the part worth noticing. Through 2025 and most of 2026 those two groups published separate answers to the same question, and a merchant deciding which one to support had no shared reference to work from. The [agentic commerce map](/blog/agentic-commerce-map-2026) lays out how far apart those answers still are.
How many venues are writing these rules?
Seven, as of August 2026, and they don't share a scope. Two are run by a single card network, two sit inside the Linux Foundation, one belongs to the FIDO Alliance, one has no foundation behind it at all, and the newest is this Alliance. No two of them govern quite the same thing.
| Venue | What it governs | Since |
|---|---|---|
| Mastercard | Agent Pay | April 2025 |
| OpenAI and Stripe (no foundation) | Agentic Commerce Protocol | September 2025 |
| Visa | Trusted Agent Protocol | October 2025 |
| Agentic AI Foundation, Linux Foundation | Model Context Protocol | December 2025 |
| FIDO Alliance | Agent Payments Protocol (AP2) | April 2026 |
| x402 Foundation, Linux Foundation | x402 | April 2026 |
| Agentic Payments Alliance | cross-rail frameworks, agent identity and authorization | August 2026 |
Three of those seven arrived in a five-month window. Google handed AP2 to the FIDO Alliance in April 2026; Coinbase handed x402 to the Linux Foundation the same month; the Agentic Commerce Protocol still has no foundation behind it at all. If you're building something that has to work in 2028, there's no way to know yet which of these your merchants will speak.
The Alliance seats both card networks and the stablecoin issuers in one room, and the early work it names is agent identity and authorization rather than settlement mechanics. That ordering is correct, and it's a better starting point than the protocol-by-protocol work that came before it. The charter itself is still to be written: Rain says members will set it together.
What can a standards body settle, and what stays yours?
A standards body settles the interface. Message formats, shared vocabulary, a common story to tell a regulator, the conformance tests that let two vendors claim they interoperate.
That work is slow and it's worth doing.
What it doesn't touch is policy. No consortium is going to decide that your research agent may spend $200 a month at data vendors and needs a person to sign off above $2,000. Those numbers came from your finance team. They change when your budget changes, and they have to be enforced by something you control.
| The Alliance can settle | You still decide |
|---|---|
| How an agent presents its identity to a merchant | Which agents you issue credentials to at all |
| A shared vocabulary for authorization decisions | The amount, the merchant category, the time window |
| Interoperability tests across card and crypto rails | Who approves before a purchase clears, and above what threshold |
| A common position for regulators | What your record has to prove afterwards, and to whom |
The right-hand column is where the money is lost, and it's the column no consortium writes for you.
What do you do while the rules are unwritten?
Choose controls that survive whichever protocol wins. A control that lives in your purchasing instrument, rather than in an integration with one protocol, keeps working when the standards settle differently than you bet, because it never depended on the standard in the first place.
An agent buying compute through x402 and an agent buying a conference ticket from a merchant that has never heard of x402 are the same budget problem with two different endpoints. Per-endpoint controls only cover the endpoints you predicted; a purchasing instrument scoped to one agent covers every merchant that accepts it, including the ones your agent finds on its own. That property doesn't depend on the Alliance publishing anything.
Shatale issues agent-scoped virtual cards with policy enforced at the authorization moment, human approval workflows above the thresholds you set, and an immutable per-agent record of every decision. A purchase outside policy is blocked or escalated for approval while the merchant is still waiting on the authorization, rather than reconciled after the statement lands. Where the rest of the category goes is the Alliance's problem to argue about.
What to ask
- Which of the seven venues does this vendor implement, and what happens to my integration if a different one wins?
- Where is my spending policy stored, and can I change it without a deploy?
- At what point in the transaction is the policy evaluated?
- If an agent tries a purchase outside policy, is it blocked, escalated to a person, or logged for me to find later?
- Who can read the record of what each agent bought, and can it be edited after the fact?
FAQ
What is the Agentic Payments Alliance?
A coalition launched by Rain on 18 August 2026 with 26 founding members, including Visa, Mastercard, Fiserv, Circle and Solana. It's run collectively by its members rather than owned by one company, and its early work covers shared research, frameworks for AI agent identity and authorization, and regulatory advocacy.
Who are the founding members?
Avalanche, Basis Theory, Chainalysis, Circle, Coinflow, Crossmint, delta Network, Episode Six, Evertec, Fireblocks, Fiserv, Kala, Lithic, Mastercard, Monad, PayOS, Rain, Remitly, Rialo by Subzero Labs, Sardine, Shift4, Solana, Turnkey, Uniswap Labs, Visa and Yuno.
Does joining an alliance mean the standards are settled?
No. The Alliance published a member list and a statement of early work in August 2026, not a specification, and its charter is still to be set by the members. Six other governance venues already cover pieces of agent payments, and the Agentic Commerce Protocol has no foundation governing it at all.
Does an agent-payments standard control how much my agent can spend?
No. Standards govern how systems exchange information about a payment. The limits, the merchant categories, the approval thresholds and the record of what happened are set and enforced by whoever runs your card programme.
---
The companion piece on [what MCP's roadmap does and doesn't cover](/blog/mcp-roadmap-agent-identity-not-payments) takes the same question to the protocol layer. Early access is free for publishers.
Shatale is the control layer for AI-agent payments. Its authorization architecture is the subject of European patent application EP26194994.5 (filed; priority 28 July 2026).