Pay-per-inference for AI agents: How BlockRun and Incarna use Amazon Bedrock AgentCore payments

Incarna, built by SpreadX, integrated Amazon Bedrock AgentCore payments so its agents can pay BlockRun for model inference one request at a time, and put the flow into production on Base. The Incarna team completed the full integration in three days, one day building and two testing, using roughly 200 lines of application code against an original scope of two to three months. During the beta, agents processed over 1,000 payments ranging from $0.001 to $0.05 per call, each settling individually on-chain.

BlockRun is a pay-as-you-go inference router serving more than 90 models from more than 15 providers over x402, with each call quoted and settled independently. AgentCore payments works with x402-compatible endpoints, including Amazon Bedrock inference endpoints. In the flow, the agent calls BlockRun, which answers with a PaymentRequired challenge and a price for that specific call; the agent opens a payment session and calls ProcessPayment, AgentCore payments checks the quote against the session's spending limits and signs the authorization from the agent's own wallet, the seller verifies the signature, and BlockRun serves the inference and records the charge.

Two x402 payment schemes are supported. The exact scheme is usually used when the price is known up front, while the upto scheme suits resources with dynamic pricing: the agent authorizes a ceiling and the provider settles for actual usage up to that ceiling. AgentCore payments is a managed capability of Amazon Bedrock AgentCore that handles the payment protocol, connects to a wallet, signs the transaction and enforces spending limits. Incarna provisions each agent's wallet with the Coinbase CDP connector; the customer owns the wallet and grants Incarna delegated authorization.

Spending ceilings are enforced at the infrastructure layer rather than by the model, so the agent's own code and prompt cannot change them even if the prompt is manipulated. Each session also carries an expiry time, and Incarna sizes sessions to a day's budget. Payments settle in a stablecoin: Incarna uses USDC on Base, and each transaction is verifiable on-chain. Setup can run through a guided conversation with the AgentCore payments skill in the Agent Toolkit for AWS in Claude Code, Kiro or Codex, or manually via the AgentCore CLI, AgentCore SDK or AWS SDK, storing Coinbase CDP or Stripe Privy credentials as a payment credential provider in AWS Secrets Manager.

Why it matters: Teams building agents that buy services at runtime can point them at x402-compatible paid endpoints such as BlockRun, and the infrastructure, not the agent, enforces the spending cap. Incarna's founders said the service supplied the customer-owned wallet, funding and revocation flow, platform-enforced limit and signing for both x402 versions, none of which they wrote.