Amazon Bedrock AgentCore Unveils First Managed Payments for AI Agents in Partnership with Coinbase and Stripe

Breaking News: AI Agents Can Now Pay Their Own Way

Amazon Web Services (AWS) today announced a groundbreaking preview of managed payment capabilities within Amazon Bedrock AgentCore, allowing AI agents to autonomously access and pay for APIs, MCP servers, web content, and even other agents. Built in collaboration with Coinbase and Stripe, the feature eliminates the need for developers to build custom billing, credential management, and compliance systems from scratch.

“This is a pivotal shift—AI agents can now handle financial transactions in real time without human intervention,” said an AWS representative. “We’re removing undifferentiated heavy lifting so builders can focus on innovation.” The integration supports Coinbase CDP wallets and Stripe Privy wallets, with session-level spending limits to maintain control.

The announcement came as part of a broader weekly update that also introduced the Agent Toolkit for AWS, a production-ready suite of tools for AI coding agents, and made the AWS MCP Server generally available. Other highlights include a preview of Amazon WorkSpaces for AI agents and new EC2 instances powered by sixth-generation Intel Xeon processors.

Background

For years, businesses using AI agents have faced a critical bottleneck: agents could gather information but couldn’t independently pay for premium data, APIs, or services. Developers had to build custom payment workflows, often leading to security gaps and compliance headaches. The new AgentCore payments feature, now in preview, directly addresses this by embedding payment infrastructure into the agent runtime.

Amazon partnered with Coinbase and Stripe—two leaders in digital payments—to create a seamless connection. Wallets from either provider can be linked to an agent, and spending limits are configured per session. During execution, the agent transacts autonomously, drawing from the wallet as needed. “It’s like giving your AI assistant a corporate credit card with built-in guardrails,” the AWS representative explained.

What This Means

This capability unlocks a new class of autonomous workflows. For example, a research agent can now pay for real-time market data while analyzing trends, or a coding agent can call paid APIs mid-task without waiting for human approval. The potential extends to multi-agent systems where one agent pays another for specialized services.

“We’re seeing early adopters already experimenting with agents that negotiate and transact autonomously,” said a cloud computing analyst. “This could fundamentally change how we think about AI-driven business processes.” The feature is available via the AgentCore CLI, with documentation and a blog post providing detailed guidance.

Agent Toolkit for AWS

Alongside the payments preview, AWS announced the Agent Toolkit for AWS—a free, production-ready collection of tools and best practices for AI coding agents. The toolkit reduces errors and token costs while enforcing enterprise-grade security controls. It succeeds the earlier MCP servers and plugins available on AWS Labs, and includes a quick start guide and a growing library of skills and plugins on GitHub.

AWS MCP Server Now Generally Available

The AWS MCP Server, a managed remote Model Context Protocol server, reached general availability. It gives AI agents and coding assistants secure, authenticated access to all AWS services through a small, fixed set of tools. Part of the Agent Toolkit, the MCP Server eliminates the need for custom integrations. Sebastian Stormacq’s blog post offers deeper technical insights.

Amazon WorkSpaces for AI Agents (Preview)

AWS also previewed Amazon WorkSpaces for AI agents, enabling agents to securely access and operate desktop applications within managed Windows environments. Organizations can automate everyday workflows at scale while maintaining full governance and compliance. Micah Walter’s blog post provides more details.

New EC2 Instances: M8idn, M8idb, R8idn, R8idb

New EC2 instance families—M8idn, M8idb, R8idn, and R8idb—are now available, powered by custom sixth-generation Intel Xeon Scalable processors and the latest AWS Nitro cards. They deliver up to 43% better compute performance per vCPU compared to previous generations. M8idn and R8idn instances offer up to 600 Gbps network bandwidth, while M8idb and R8idb instances provide up to 300 Gbps EBS bandwidth.

Valkey Turns Two: Open Source Milestone

In related news, the Valkey open-source project celebrated its second anniversary, surpassing 100 million Docker pulls—a 17x year-over-year increase. The project now counts over 225 contributors. “Valkey proves that open, community-driven technology innovates faster and scales further than any single-vendor model,” an AWS statement noted.

For a complete list of AWS announcements, visit the What’s New with AWS page.