Resources for operationalizing agentic AI - AWS Prescriptive Guidance

Resources for operationalizing agentic AI

AWS services

The following AWS services and features can help you build and operationalize agentic AI systems in the AWS Cloud:

  • Amazon API Gateway can expose agent capabilities as scalable and offers usage-based pricing.

  • AWS AppConfig offers runtime configuration management and feature toggling for agents across tenants or environments.

  • Amazon Bedrock is a foundation model service that agents can use for reasoning, generation, and prompt execution.

  • AWS Cloud Development Kit (AWS CDK) is an infrastructure as code service that you can use to deploy and manage agent stacks.

  • AWS CloudTrail records event history so that you can track agent activity, audit trails, and integration behaviors.

  • Amazon CloudWatch can manage logs, metrics, and alarms for monitoring agent performance and multi-agent collaboration behavior.

  • AWS CodePipeline provides CI/CD automation that you can use to test, validate, and deploy agent code.

  • Amazon Cognito is an identity service that you can use to manage user and tenant authentication in multi-agent systems.

  • AWS Config offers compliance and drift detection for agent policy and environment configuration.

  • AWS Cost Explorer can track agent-level usage and help align costs to maximize your ROI.

  • Amazon DynamoDB is a storage service that you can use for agent memory, improvement logs, and contextual state.

  • Amazon Elastic File System (Amazon EFS) is a shared file system that you can use for agent collaboration or intermediate processing across workflows.

  • Amazon EventBridge is a core event bus that you can use to route tasks and orchestrate communication in the agent fabric.

  • Amazon EventBridge Pipes can streamline event ingestion and routing for connecting agents and services.

  • Amazon GuardDuty offers threat detection and anomaly monitoring that can support safe agent execution.

  • AWS Identity and Access Management (IAM) helps you define fine-grained permissions for agent execution and data access.

  • AWS Lambda is a stateless compute service that can run agent logic and swarm drones.

  • AWS Marketplace is an external distribution platform that you can use to offer agent capabilities as commercial products.

  • AWS Organizations is a cross-account governance and policy enforcement service that can help you manage multi-tenant agent infrastructure.

  • AWS Organizations service control policies act as guardrails for controlling permissions at the account or organizational unit level.

  • Amazon QuickSight is a visualization tool that you can use to track KPIs, decision outcomes, and agent impact.

  • AWS Resource Access Manager (AWS RAM) can help you share capabilities between accounts and agent services.

  • Amazon SageMaker AI is a service that you can use for model training, fine-tuning, and inference beyond foundational models.

  • Amazon Simple Storage Service (Amazon S3) offers object storage for prompt libraries, model artifacts, and agent-generated data.

  • AWS Step Functions is a workflow engine that can help you coordinate multi-agent flows and retraining pipelines.

  • AWS X-Ray offers distributed tracing that you can use to track agent decision flows and service dependencies.

Other AWS resources