n8n Adds Amazon Bedrock AgentCore for Stateful Agents on AWS.

n8n is bringing more of Amazon’s managed AI agent infrastructure directly onto its workflow canvas. The new Amazon Bedrock AgentCore node gives builders access to persistent customer memory, sandboxed code execution and a managed cloud browser, with the underlying services running inside their own AWS account.

The company announced the canvas integration on LinkedIn, alongside templates intended to shorten the setup process. The central pitch is consolidation: teams can configure several pieces of agent infrastructure through one node rather than assembling separate memory, browser and execution systems around a workflow.

That could make a meaningful difference for technical teams moving agents beyond prototypes. A basic chatbot can often operate without remembering previous interactions or taking actions in external systems. Production agents usually need both, while also requiring controls around code execution, browser access and customer data.

Amazon Bedrock AgentCore gives n8n workflows access to three important capabilities:

• Persistent, customer-scoped memory: Agents can retain context for an individual customer across interactions, reducing the need for users to repeat information.
• Sandboxed code interpretation: An agent can run code in an isolated environment, useful for calculations, data processing and other tasks that should not execute directly within the main workflow environment.
• Managed cloud browsing: Agents can investigate websites or interact with browser-based services without teams maintaining their own browser automation infrastructure.

According to n8n, these components operate in the user’s AWS account and are configured from a single canvas node. The company’s technical blog post demonstrates the integration through a customer-support workflow built around multiple specialist agents.

In that example, a triage agent sends each request to one of three specialists. One runs calculations in the sandbox, another retrieves AWS guidance, and a third conducts research. All four agents use the same AgentCore harness and share memory associated with the customer, allowing work to move between specialists without repeatedly gathering the same context.

The architecture is notable because multi-agent workflows can become difficult to coordinate quickly. Each specialist may require access to the same conversation history, but not necessarily the memory belonging to another customer. Customer-scoped persistence gives builders a defined way to preserve continuity while separating sessions.

Keeping the AgentCore services inside a company’s AWS environment can offer more control over deployment, access policies and operational oversight than routing those functions through an unrelated third-party stack. It may be particularly relevant for teams already standardizing infrastructure and governance around AWS.

It does not remove the need for careful configuration. Builders still need to decide what should enter persistent memory, which tools an agent can call, and when browser or code execution is appropriate. A sandbox reduces the exposure created by generated code, but workflow permissions and data-handling rules remain important parts of deployment.

For small businesses, the integration is most practical where an agent needs to complete multi-step work rather than simply produce text. Potential applications include support investigation, account-specific research, numerical analysis and workflows that combine web browsing with internal data. The accompanying templates should provide a faster starting point, although n8n has not disclosed specific pricing or plan requirements in the supplied announcement.

The release expands on n8n’s earlier work around stateful agents, covered in n8n Adds Amazon Bedrock AgentCore for Stateful Multi-Agent Support. The canvas node now makes that infrastructure more directly accessible within the same interface used to design the rest of an automation.

What is n8n Amazon Bedrock AgentCore?

It is an n8n canvas integration for configuring Amazon Bedrock AgentCore services within an automation workflow. It provides customer-scoped persistent memory, sandboxed code interpretation and managed cloud browsing, with the services running in the builder’s AWS account.

When is Amazon Bedrock AgentCore available in n8n?

n8n said the node was available from the canvas in its August 25, 2026 announcement. The company also released templates to help builders get started, but the supplied sources do not specify plan restrictions or pricing.

Who is the integration designed for?

It is aimed at teams building stateful AI agents that need to remember customers, execute code safely or perform browser-based tasks. It is especially relevant to organizations already using AWS and to multi-agent workflows where several specialist agents need shared, customer-specific context.