--- title: Configure an IAM for Agent discovery description: Configuring IAM policy action permissions necessary for Agent discovery and integration user access on AWS to read bedrock agentic data. locale: en-US canonical_url: https://www.servicenow.com/docs/r/australia/intelligent-experiences/ai-control-tower/configure-iam-for-agent-discovery.html release: australia product: AI Control Tower classification: ai-control-tower topic_type: task last_updated: "2026-09-10" reading_time_minutes: 1 keywords: [Now Assist, generative AI] breadcrumb: [Configure AWS console, Hyperscaler connection with AWS, Using AI Control Tower to create hyperscaler connections, Using AI Control Tower \(legacy\), AI Control Tower \(legacy\), Establishing AI governance, Enable AI Experiences] --- # Configure an IAM for Agent discovery Configuring IAM policy action permissions necessary for Agent discovery and integration user access on AWS to read bedrock agentic data. ## Before you begin Role required: Admin AWS provides the minimum given policies by default: - AmazonBedrockReadOnly - CloudWatchReadOnlyAccess The minimum policy action permissions needed for the integration user on AWS to read bedrock agentic data: - ListAgents - GetAgent - ListAgentActionsGroups - ListAgentVersions - GetInferenceProfile - GetFoundationModel - ListAgentCollaborators - GetAgentAlias **Note:** To exclude specific resources from AI Control Tower, you can restrict them from the resources region. Select 'all' to include all AWS bedrock agentic resources data into AI Control Tower. ``` { "Version": "2012-10-17", "Statement": [ { "Sid": "VisualEditor0", "Effect": "Allow", "Action": [ "bedrock:ListAgents", "bedrock:ListAgentVersions", "bedrock:ListAgentCollaborators", "bedrock:ListAgentActionGroups", "bedrock:GetInferenceProfile", "bedrock:GetFoundationModel", "bedrock:GetAgent" ], "Resource": "*" } ] } ``` ## What to do next Configure an IAM policy for Execution monitoring.