How IT must change: moving from manual ops to autonomous, AI-driven infrastructure

Modern enterprise infrastructure is changing faster than traditional IT operating models can manage. Annual or semi-annual update cycles and manual intervention paradigms were designed for relatively static stacks. The mass adoption of AI and increasingly complex, multi-domain environments have exposed a condition Cisco calls operational obsolescence: operations processes that are inadequate for the pace, scale, and risk profile of current architectures.

The three technical pillars for modernization

• Make desired infrastructure state explicit, versionable, and managed through software development lifecycle practices.
• Treat network and infrastructure elements as software artifacts that can be reviewed, tested, and rolled back.

• Use machine-speed automation to execute, remediate, and maintain consistency across environments at a scale human teams cannot match.
• Move routine and repeatable execution to deterministic, observable automation.

• Integrate governance, approvals, and change traceability into the code lifecycle so every change is verified against security and operational policies.
• Embed policy-as-code and automated testing into pipelines to reduce drift and manual errors.

Cisco's framing includes a descriptive line: "With Cisco® Services as Code, enterprises can define their network infrastructure state and treat all network elements as software that can be versioned and managed at scale."

Operational foundations required before scaling autonomy

• Establish a single Source of Truth (SoT), version-controlled data models, golden configurations, design baselines, and Method of Procedures (MOPs).
• Make infrastructure intent machine-readable to prevent configuration drift across multi-domain environments.

Automation & Guardrails (the boundaries)

• Build programmable workflows with deterministic boundaries for autonomous execution.
• Implement pre-change simulations and policy-as-code guardrails to prevent unsafe actions.

• Retain HITL gates for high-consequence changes, ensuring human judgment for exceptions and complex risk decisions.
• Integrate approvals and traceability directly into automated pipelines.

Continuous validation and observability

• Validate changes in real time and maintain audit trails so remediation and rollback can be automated and verifiable.
• Use observability to confirm that automated actions produce intended outcomes and to detect emergent failure modes.

Practical implications for IT leaders

• Define which operations remain human-controlled and codify those decision points as governance policies.

The recommended shift is incremental: codify intent, automate within defined boundaries, and expand autonomous execution only after establishing rigorous validation and HITL controls. This sequence reduces operational risk while enabling faster, safer innovation.