Ten Agentic Payment Protocols Launched in Six Months. Here's What That Actually Means.
Between October 2025 and April 2026, more than ten payment protocols launched to let AI agents spend money on their own. x402, AP2, MPP, L402, Visa TAP, Mastercard Agent Pay, Amex ACE, Google/Shopify UCP, OpenAI ACP — the list keeps growing. Fenwick & West called it in April: 2026 is the year of agentic payments.
They're not wrong. But the protocol race is solving the wrong problem first.
What Agentic Payments Actually Need
Every one of those protocols solves the mechanics of payment: how an agent authenticates, how a transaction clears, how a stablecoin or card gets charged. What none of them ship with is the governance layer: what an agent is allowed to spend, against which counterparties, under what conditions, with what audit trail, and who can revoke access when something goes wrong.
The result is that enterprises can now technically give an agent a wallet. What they can't do yet is give that agent a wallet they're comfortable with from a compliance standpoint.
The Credit-Check Layer Nobody Built (Until Now)
Think about what happens to a human payment before it clears: KYC, AML screening, sanctions check, fraud scoring, velocity rules, and a paper trail. An AI agent running on x402 or AP2 today gets none of that. The API key works, the USDC clears, the transaction posts — and nobody asked whether that counterparty was on a sanctions list.
Agent FICO (300-850) is the behavioral trust score we built to close that gap. Every agent builds a reputation score based on spend patterns, counterparty history, and policy adherence — the same way a credit bureau builds a human credit score, except in real time and keyed to cryptographic agent identity (ECDSA P-256, not bearer tokens).
MnemoPay layers on top of whichever payment protocol your stack already uses. You don't have to pick one protocol and bet on it winning — you add the governance layer once, and it works regardless of what clears underneath.
Why Protocol Count Doesn't Matter
The reason ten protocols launched in six months is the same reason ten competing standards emerge whenever a new infrastructure primitive appears: everyone wants to own the rails. Most of them won't survive. The ones that do will converge on the same clearing mechanics they always converge on.
What survives in every infrastructure race isn't the payment rail — it's the trust and compliance layer that sits above it. SWIFT isn't interesting because of its messaging format. It's interesting because of the correspondent banking relationships and sanctions screening built on top.
MnemoPay is the correspondent-banking layer for agent payments: 672+ tests, v1.0.0-beta.1 shipped, 1.4K weekly npm downloads. Protocol-agnostic by design.
NOTE: source is a Fenwick & West legal publication (source=other, no comment section). Rerouting from article to owned Dev.to channel. Verify before publishing.
Top comments (0)