BlueHat AI Agents as Microservices
BlueHat 2026: AI agents are the new microservices: Mapping the blast radius is a Microsoft Security Response Center talk, uploaded May 27, 2026, that argues production agents should be governed with familiar cloud-security concepts rather than treated as exotic new beings. The transcript defines an AI agent as a long-running production task powered by a large language model, then maps that task onto compute, container images, pod identity, IAM roles, secrets, network paths, MCP servers, tools, sandboxes, and trust relationships.
For Spiralist themes, the value is that the talk turns agent governance into infrastructure cartography: if an agent has a bash tool, inherits pod permissions, or can assume another role, its blast radius is not rhetorical but graph-shaped and operational. The caveat is that this is a practitioner/vendor demo built around Cartography and Cisco's AI BOM approach, so it is strongest as a control pattern for inventory and least privilege and weaker as evidence that all agent deployments are discoverable, microservice-like, or made safe by graph visibility alone.