Best future architecture AIPP/ASF perform the action. ITSM records, authorizes, scores, and governs the action.
Example flow: Monitor detects production issue. ASF routes to diagnostic agents. AIPP agent proposes fix, test, rollback, release note. Policy engine classifies risk. Low-risk fix auto-executes. High-risk fix opens change approval. ITSM records incident, affected service, timeline, evidence, approval, root cause, and postmortem. Knowledge base updates automatically.
That is not “old ITSM.” That is agentic ITSM.
The dangerous version without ITSM Without ITSM, AIPP + ASF can become: A swarm of smart agents with no durable accountability. Autonomous remediation with weak approval trails. Beautiful code changes with unclear business ownership. Duplicate fixes across teams. No single source of truth for incidents, changes, SLAs, or service impact. No defensible audit history when something fails. For healthcare, enterprise AI, model factories, and production inference, that is too risky.
My blunt take You do not need ITSM as a bureaucratic ticket jail. You do need ITSM as the service truth layer.
The winning pattern is: ITSM becomes the ledger. ASF becomes the nervous system. AIPP becomes the builder/healer. Humans become policy, design, ethics, escalation, and final accountability. So: keep ITSM, but refactor it into an AI-native Service Governance Fabric.