aws AWS Open Source Blog ·

AWS introduces Loom for secure AI agent development and deployment

blogaiawsengineer
feature announcement

AWS has released Loom, an enterprise-grade platform designed to simplify the secure development, deployment, and operation of AI agents using Amazon Bedrock AgentCore. This addresses challenges organizations face in implementing security and governance frameworks for agentic capabilities. Loom offers a unified management UI, identity provider integration, and built-in security best practices, targeting platform engineering teams looking to build AI applications on AWS.

  • Loom platform for building and deploying AI agents
  • Automated resource tagging for governance and cost attribution
  • Role-based and attribute-based access control for agent security
  • Secure agent deployment with paved path blueprints
  • Loom's core tenets: simple, opinionated, enterprise-grade
Features (1)
  • Loom platform for building and deploying AI agents

    Loom is an enterprise-grade platform released by AWS Labs that provides a unified management UI and backend API for building, deploying, and operating AI agents. It integrates with AWS Strands Agents and Amazon Bedrock AgentCore Runtime, simplifying agent creation with options for low-code and no-code deployment.

Enhancements (3)
  • Automated resource tagging for governance and cost attribution

    Loom enforces consistent application of tags on deployed resources, including required tags like 'loom:application', 'loom:group', and 'loom:owner', and allows for custom tags. This aids in cost attribution and the enforcement of conditional governance policies.

  • Role-based and attribute-based access control for agent security

    The platform implements fine-grained access control through a combination of role types (administrator, end user) and group tags. This approach, utilizing RBAC and ABAC, ensures users have appropriate access to specific agent resources and data within multi-tenant environments.

  • Secure agent deployment with paved path blueprints

    Loom leverages paved path blueprints, which are pre-validated configurations incorporating best practices for security, least privilege, and integration with cost management. This enables platform engineering teams to rigorously test and deploy agentic applications.

Notes (1)
  • Loom's core tenets: simple, opinionated, enterprise-grade

    Loom is designed to be simple for builders, opinionated by using Strands Agents SDK and AWS managed services, and enterprise-grade by implementing best practices for security and governance through a paved path approach.

Read the original announcement →

https://aws.amazon.com/blogs/opensource/building-secure-ai-agents-at-scale-introducing-loom-for-aws/

Related releases