
What happened
Incarna used Amazon Bedrock AgentCore payments to pay BlockRun per inference over x402, completing the integration in three days instead of the two to three months originally scoped, with over 1,000 payments from $0.001 to $0.05 each.
Why it matters
Spending limits are enforced by the infrastructure rather than the model, so an autonomous agent cannot exceed the ceiling even if its prompt is manipulated, which may make it more comfortable for builders to let agents move real money.
WHO IT HITSTeams building AI agents that need to buy services mid-task, such as inference or API calls, can add payment support through one managed integration instead of assembling wallets, signing, and spending controls themselves.
Summaries like this, in your inbox every morning.
AI agents frequently need to buy small things while they work, such as a model inference, an API response, or access to web content. These purchases happen inside the agent's loop with no person available to approve them, and each can be worth a fraction of a cent, a pattern card networks were not built for. Amazon Bedrock AgentCore payments addresses this by handling the protocol, connecting to a wallet, signing the transaction, and enforcing spending limits, rather than making builders solve each part separately.
In this integration, BlockRun is the seller, serving metered inference from a live catalog of more than 90 models from more than 15 providers over x402, with each call quoted, paid, and settled independently. When a paid endpoint answers with HTTP 402, the agent uses AgentCore payments to pay over x402, signing from the agent's own wallet address. Settlement happens in a stablecoin, and Incarna uses USDC on the Base network, with each transaction verifiable on-chain. That makes the on-chain payer the agent's own identity rather than a shared platform key, so every transaction is traceable to a single durable entity.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
CoreWeave launched CoreWeave Forge, connecting serving, observability, post-training and evaluation
Samsung Electronics' next-generation 12-layer HBM4E high-bandwidth memory has reportedly cleared customer qual…

Kelsey Kuan, general manager of HD Renewable Energy subsidiary Star Trade, told DIGITIMES that energy storage…

Stanford and Carnegie Mellon researchers tracked 1182 Character.AI users from September 2024 to August 2025, t…

A study presented at USENIX Security Symposium 2026 examined 196,682 resumes from hiring platforms and detecte…

Google Cloud announced Gemini Agent, a cloud-based multi-agent tool
