DNA Insider
繁中
Back
Official Source· AWS Artificial Intelligence· AI· Published Sep 30, 2026

AWS Outlines a Multi-Account AI Agent Using AgentCore Gateway and MCP

In a September 24, 2026 technical guide, AWS described how an AI agent can query data across AWS accounts through Amazon Bedrock AgentCore Gateway while the data stays in the account that owns it.

AWS Outlines a Multi-Account AI Agent Using AgentCore Gateway and MCP

Full Report

AWS published “Build a multi-account AI agent with AgentCore Gateway and MCP” on September 24, 2026, outlining an architecture for agents that work across AWS accounts. A central platform account runs the agent, while line-of-business accounts expose data and tools through MCP servers. Amazon Bedrock AgentCore Gateway brings those servers together behind one endpoint for tool discovery and invocation.

The design keeps source data in its owning line-of-business account. When a query is made, an MCP server runs the requested tool locally and returns its result to the platform account rather than moving the underlying dataset. The guide also describes cross-account authentication and authorization, including OAuth 2.0 machine-to-machine credentials managed through AgentCore Identity and fine-grained controls using Policy in AgentCore; the walkthrough uses Okta.

AWS provides an implementation that bootstraps resources across four accounts using AWS CDK, including the platform and line-of-business resources, MCP servers, Gateway targets, and a sample web application. The guide also covers continuous evaluation, version rollouts, and observability. In a related example published September 29, AWS described a contract-intelligence platform in which agents extract and verify contract fields and Amazon Quick analytics answers questions about individual contracts and the portfolio.

Why This Matters

The architecture separates cross-team querying from centralizing the underlying data: line-of-business teams retain their data and tool boundaries, while the platform team can manage agent access through a shared Gateway and authorization policies. The implementation guide gives readers a concrete reference for the identity, access-control, and operational elements involved in building that pattern.

Evidence

Start with the primary evidence, then review supporting sources and context.

1 items

Primary Evidence

1
  • OfficialAWS Artificial IntelligenceDocumentOriginal

    Build a multi-account AI agent with AgentCore Gateway and MCP

    Senthil Kamala Rathinam

    “Build a multi-account architecture that keeps each team's data in its own AWS account while giving AI agents a unified way to query across them. A central platform account runs the agent using Amazon Bedrock AgentCore Gateway and MCP, while line-of-business accounts expose their data as MCP servers with secure cross-account access and fine-grained…”

Recommended signals

Recommended by brand, topic, and recent related coverage

No more recommended signals right now

Official statement

Open original source
Back to Live FeedExplore more signals