← System map

Domain 08 / 10

IT & Agentic Operations

IT shifts from managing applications to orchestrating fleets of AI agents — a new function of agentic controllers, model monitoring, and AI-specific security.

How it works

  1. 1Agent orchestration layer coordinates AI agents across clinical and admin domains
  2. 2Agentic controllers supervise, override, and audit agent behavior
  3. 3Continuous model monitoring for drift, bias, and performance degradation
  4. 4AI-specific security: prompt injection defense, model access control, adversarial monitoring

Autonomy today

Most health systems have no dedicated agentic-ops function yet; this role is nascent industry-wide, not just in healthcare.

In ~5 years

Agentic operations is a standard IT department function, as fundamental as network security is today.

Flaws & risks

  • An automated system is an attackable system — agentic infrastructure is a new, poorly understood attack surface
  • Talent for this role barely exists yet; it will need to be grown internally or hired from adjacent fields

Who's building it

Free preview · full directory in Pro
Credo AIemergingAI governance/monitoring

AI governance platform for model risk and compliance monitoring.

Protect AIemergingAI security

Security tooling for ML/AI systems and supply chain.

Preparation checklist

  • Create the agentic-controller role now, even at small scale, to build institutional knowledge early
  • Treat AI system security as its own budget line, not a subset of general IT security