<aside> 🧭
Bottom line: Agentic payments are not one market. They are a stack that begins with probabilistic intent and ends in deterministic settlement. Crypto is strongest where software pays software, where transactions must be programmable, and where value must move globally at machine speed. The likely winners are specialists in agent runtimes, transaction execution, wallets and authorization, protocol coordination, stablecoin infrastructure, and risk—not one provider owning everything.
</aside>
drafed X post
Agentic Payments Are Not One Market
Agentic payments are payments that software agents can initiate, authorize within delegated limits, execute, and reconcile on behalf of people, companies, or other agents. The category overlaps with agentic commerce, but it is broader: it includes API micropayments, autonomous procurement, DeFi execution, treasury operations, cross-border routing, payouts, and agent-to-agent service markets.
The most important architectural fact is the separation between probabilistic reasoning and deterministic financial authority. An agent may search, negotiate, choose a route, or construct a transaction, but an independent authorization boundary must determine whether it can spend, and a deterministic rail must settle without reinterpretation. The IMF uses a three-layer model—intent and orchestration, control and authorization, and settlement—which this report expands into seven commercial layers plus a cross-cutting trust layer (IMF, April 2026).
Crypto has a structural advantage in machine-to-machine commerce because wallets are programmatic accounts, stablecoins provide internet-native value, and blockchains provide global, always-on settlement. The strongest live wedge is pay-per-use digital services: APIs, model inference, data, compute, browser sessions, storage, and other resources that agents consume mid-task. x402 and MPP are competing or complementary ways to coordinate these payments. Crypto-native financial execution—swaps, bridges, lending, staking, perps, and prediction markets—is also live, but carries materially higher execution and authorization risk.
The market is real but early. Chainalysis found more than 100 million cumulative x402 transactions on Base through Q1 2026, but much of the spike was driven by speculative pay-to-mint activity. It concludes that x402 has moved beyond proof of concept while mass adoption remains distant and current participants remain crypto-native (Chainalysis). Solana separately reports more than 35 million x402 transactions and over $10 million in volume since launch (Solana). These figures prove technical use and early demand, not yet a broad autonomous economy.
For Aomi, the correct category is onchain agent execution runtime. Aomi belongs above wallets and settlement rails: it hosts the agent loop and tools, turns intent into typed actions and transactions, simulates outcomes, manages state, and hands a verified request to an external signer. Payment protocols such as x402 are capabilities that Aomi can use; they are not Aomi's identity.
Research date: August 11, 2026.
Included: consumer and enterprise agentic payments, crypto-native agent wallets, stablecoin and card rails, machine-payment protocols, agent frameworks, DeFi execution, service discovery, compliance, and transaction security.
Excluded: conventional AI fraud models that do not participate in an agent-initiated payment flow; generic crypto infrastructure with no meaningful agent-facing product; purely speculative tokens whose only connection is an “AI agent” label.
Evidence hierarchy: