Anthropic adds AWS authentication and billing support for Claude Managed Agents

Anthropic’s agent platform just got a lot easier for big companies to buy and govern. On May 11, Anthropic and AWS launched Claude Platform on AWS, which lets customers use Anthropic’s native tools — including Claude Managed Agents — through their existing AWS account. That means AWS IAM for sign-in, AWS billing for spend, and CloudTrail for audit logs. The gap this closes is pretty obvious: enterprises like frontier AI tools, but they hate separate credentials, separate invoices, and separate oversight. ### What actually changed? The news is not just “Anthropic likes AWS.” It’s that Anthropic’s own platform is now available through AWS as a generally available service, and Managed Agents are part of that package. AWS says customers get the same native Claude Platform experience through their AWS account, without separate credentials, contracts, or billing relationships. Anthropic says this includes day-one access to the full platform and new betas as they ship. (claude.com) ### Why does IAM matter so much? Because identity is where enterprise rollouts usually get stuck. Instead of handing out a new Anthropic login or API key, teams can use existing AWS IAM credentials and policies. Access can be scoped to workspaces with IAM policies, and the Claude Console sign-in is federated through AWS. Basically, security teams can treat this more like another AWS service than a separate SaaS island. (claude.com) ### What about billing? This is the other big unlock. Usage is billed through AWS Marketplace on a consumption basis, and Anthropic says it lands on a single AWS invoice and can retire against existing AWS commitments. That matters for chargebacks, budgeting, and procurement — especially in companies where getting a new vendor approved takes longer than the technical work. The Claude console also shows cost breakdowns, while AWS remains the billing relationship. (aws.amazon.com) ### So where do Managed Agents fit? Managed Agents are Anthropic’s hosted service for long-running agent work. Anthropic introduced them in April as a way to run long-horizon agents through stable abstractions like sessions, harnesses, and sandboxes. Now those agents can be deployed through the AWS-wrapped version of Claude Platform, alongside code execution, Agent Skills, files, memory stores, and other native tools. In plain English — the “brain” stays Anthropic’s, but the front door now speaks AWS. (claude.com) ### Is this the same as Claude on Bedrock? No — and that distinction is the catch. Claude Platform on AWS gives access to Anthropic’s first-party platform, but the service is operated by Anthropic and customer data is processed outside the AWS security boundary. Bedrock is the opposite tradeoff: it keeps data within AWS infrastructure and wraps Claude inside AWS-managed services like Guardrails and Knowledge Bases. So this launch is about native Anthropic features with AWS controls, not about moving Anthropic fully inside AWS. (aws.amazon.com) ### What does the console look like in practice? The AWS-linked Claude Console exposes pages for agents, sessions, environments, credential vaults, memory stores, files, and skills. But there are a few wrinkles. Organization switching is limited, billing itself is still handled through AWS Marketplace, and IAM — not Anthropic’s own member system — governs access. So it feels familiar to Claude users, but the control plane is clearly bent around AWS account structure. (aws.amazon.com) ### Why does this matter now? Because agent pilots are running into boring enterprise problems before they hit model limits. Companies can tolerate impressive demos with separate logins. They struggle to productionize them when security, finance, and audit teams get involved. Anthropic’s move makes Managed Agents easier to slot into existing AWS operating models without giving up the native platform. That is less flashy than a new model launch — but for enterprise adoption, it may matter more. (docs.aws.amazon.com) ### Bottom line? Anthropic did not just add a convenience feature. It gave Claude Managed Agents an AWS-shaped wrapper for identity, billing, and oversight. That lowers a real deployment barrier for enterprises — even if the deeper tradeoff remains the same: native Anthropic features in exchange for data processing outside AWS’s boundary. (aws.amazon.com) (claude.com)