Zerion has connected its API to AgentCash, a middleware toolkit that lets AI agents pay for data access using stablecoins instead of traditional API keys or subscription plans. The setup charges approximately $0.01 in USDC per API call, which means an AI agent can pull wallet balances, transaction histories, and DeFi position data across more than 8,000 protocols for about the cost of a single gumball.

The integration supports two payment rails: the x402 protocol running on Base and the Machine Payment Protocol (MPP) settling on Tempo. Together, they remove the conventional onboarding process that typically involves signing up, generating keys, and managing subscription tiers.

How the plumbing works

Think of x402 like a digital toll booth. When an AI agent requests data from Zerion’s API, the server responds with an HTTP 402 status code, the rarely used “Payment Required” response that has been part of the HTTP specification since the early 1990s. The agent’s wallet then sends USDC to complete the transaction, and the data flows back.

AgentCash sits in the middle of this exchange, handling wallet management, authentication with premium APIs, and payment execution in a single unified process. It launched in early July 2026 as a command-line toolkit designed specifically for machine-to-machine commerce.

The toolkit initially supported around 250 premium APIs at launch. That number has since grown to over 3,200 APIs, with compatibility across more than 500 applications, including platforms like Coinbase Wallet.

Zerion’s API covers enriched data across 38+ blockchain networks, spanning both EVM-compatible chains and Solana. That means an AI agent can query profit-and-loss data, token balances, and transaction histories across a broad swath of the crypto ecosystem without needing separate integrations for each chain.

The timeline and technical stack

The pieces of this integration came together over several months in 2026. Zerion added x402 protocol support in mid-March, which enabled the basic pay-per-call functionality on Base. MPP integration followed in late April, adding a second settlement layer through Tempo. AgentCash then launched in early July, tying everything together into a toolkit that agents can use without human intervention.

The sequencing matters because each layer addresses a different problem. The x402 protocol handles the request-response payment cycle. MPP provides an alternative settlement mechanism. AgentCash wraps both into something an AI agent can operate autonomously, managing its own wallet and deciding when and how to pay for data. For developers building agent workflows, that lowers friction around accessing premium data while keeping payment tied directly to usage rather than to a pre-set account tier.

Why micropayments for machines are gaining traction

At roughly $0.01 per call, an AI agent running a portfolio monitoring service that checks positions every few minutes would spend a few dollars per day on data access. Compared with traditional financial data terminal subscriptions that can cost thousands of dollars monthly, the cost structure is very different.

AgentCash’s expansion from 250 to more than 3,200 supported APIs in a relatively short period suggests that other data providers beyond Zerion see value in this model. The toolkit is positioning itself as a universal payment layer for machine-to-machine API commerce, with the practical test now being whether more services adopt the same pay-per-call flow across additional apps and chains.