Crypto giant Coinbase has teamed up with Amazon Web Services (AWS) to roll out a new payments layer that allows AI agents to make autonomous micropayments using the stablecoin USDC.
The integration, announced on Thursday, combines Coinbase’s wallet infrastructure and its x402 protocol with AWS’s AgentCore Payments system inside Amazon Bedrock.
The deal is taking place amid growing integration between traditional traders and AI agents, where corporations are investing in AI technologies that can both reason and take action, as well as pay for their requirements.
The technology is meant to bridge an existing loophole in the way AI systems function today. The current AI agents can analyze data, develop code, process information, and undertake complex activities.
However, when it comes to costs, both internal and external, they cannot do without human input in many respects.The deal aims at resolving the need for human input in terms of AI tools.
New system lets developers set spending limits for AI agents
The new system is designed to change conventional AI agents’ activity by allowing the agents to operate with assigned budgets and clear permissions. Developers can define how much an agent is allowed to spend, and under what conditions. This can be done via stop limits, budget limits etc.
Once the parameters are set, the agent can independently pay for services such as APIs, data feeds, or cloud infrastructure using USDC.
Payments will be made using on-chain methods with fast settlement through networks such as Base and Solana.
The key component of the protocol is the x402 protocol, which is an open-source protocol designed by Coinbase using the HTTP 402 “Payment Required” status code. In simple terms, it allows websites and services to request payment directly from machines.
Instead of stopping at a paywall designed for humans, the system enables automated agents to respond, pay, and continue their task without interruption.
New setup to help with speed
One of the key benefits of the entire system is speed and economy. The time for settling of transactions is less than 200 milliseconds, while the fees charged from the user side is only cents per transaction, hence allowing for the feasibility of micropayments.
As a result, this will facilitate AI agents’ ability to make several payments at times when they need access to certain services or data.
The developer’s view of this process would mean that the system has been created in a manner that enables users to pay using a single API call.
This is because the wallet authentication and settlement of transactions take place behind the scenes. The AI agents cannot own private keys since they cannot manage the funds. The design also includes enterprise-focused safeguards that will allow firms to set strict controls around spending, such as daily or time-based budgets that limit how much an agent can use.
AWS and Coinbase have also announced that the firms have built in compliance and monitoring features. Every transaction made by an AI agent can be tracked, logged, and reviewed through dashboards and audit systems.
The level of visibility is intended to address regulatory and operational concerns that have traditionally slowed down the adoption of autonomous payment technologies in enterprise environments.
From a broader perspective, the initiative indicates a trend towards an increasingly discussed phenomenon known as “agentic commerce.” The idea here is that rather than AI serving merely as a tool reacting to commands, it becomes an integral part of digital economies, able to explore available services on its own, analyze prices, and even conduct transactions independently of any human command.
With a combination of artificial intelligence technology and payments in a stablecoin form, such as Coinbase’s USDC and Amazon Web Services, it appears that companies aim to enable digital dollar settlement between machines.
