AI agents are moving into live operations at scale, exposing a sharp gap between executive ambition and organizational readiness. Kyndryl's Readiness Report found that 87% of business leaders expect AI to reshape career paths and role responsibilities, yet only 29% said their staff can use AI effectively and 62% remain in the experimentation phase.
The shift forces operations teams to plan how work is assigned, decisions are governed, and human accountability is maintained as autonomous systems take on more real-world actions. "Success with AI agents," said Oana Beattie, VP Data and AI at Kyndryl UK&I, "will depend less on capability in isolation and more on whether the enterprise is properly positioned to govern, orchestrate and operationalize them."
That means alignment between business intent, decision rights, data access, sovereignty, workflow design, governance and human oversight. Without that foundation, agents will struggle to deliver value at scale and may introduce fresh risks. This is a key reason the conversation around AI Agents & Automation now centers on operational readiness, not just technical capability.
Operational complexity in practice
Agentic systems are not generative AI with a sophisticated front end. They can plan, take action, coordinate tasks and move across multi-step workflows with limited human input. This makes them more capable but materially harder to govern. Once technology can reason, invoke tools and coordinate across systems, the question is no longer simply what it can do but whether the organization is equipped to run it.
Smaller agentic deployments usually rely on bounded data sets and touch few systems. At scale, more integrations, more operational variation and greater governance demands all add pressure. Many pilots lose momentum at this point because scaling AI forces organizations to confront longstanding architectural and organizational debt.
Real enterprise deployments depend on orchestration, tools, context, memory, workflow logic, policy, permissions, runtime controls, identity management, observability and human escalation paths working together. If the discussion collapses into an LLM discussion, teams risk under-designing the components that determine whether a system can work safely in production.
Scaling appropriately
As agents scale, they place pressure on enterprises in four ways. Data strain increases as sensitive and unstructured information becomes more widely available and exposed. Integration strain grows because each additional agent increases dependence on existing platforms and interfaces. Operational strain compounds as more autonomous components interact, multiplying failure scenarios. Governance strain appears when oversight models designed for static systems struggle to keep pace with dynamic behavior.
Managing this well depends on a few practical disciplines. Set clear decision boundaries that establish what agents can decide, what must be escalated and what remains under human control. Design orchestration for scale with coordinated workflows, shared context and clear control points. Build intervention into the operating model from the outset through thresholds, alerts, approvals and kill-switch mechanisms. Assign accountability to named roles and systems of record so every decision can be traced, challenged and defended.
Control must move much closer to the runtime. Policy cannot sit in a document or governance forum. It needs to be machine-readable, testable and enforceable. Permissions should flex with context, and escalation must be built from the start.
Core operational requirements
Telemetry, orchestration, real-time monitoring and AIOps are now core operational requirements. Visibility must go beyond uptime and response times to include behavior, alignment with intent, workflow dependencies, exception trends and policy adherence. Testing also needs to evolve. If a system is dynamic and sensitive to context, teams cannot test it as if it were a deterministic workflow. They are testing behavior, not just answers.
This creates a difficult leadership balance. Too much autonomy without sufficient control creates unmanaged risk. Too much control without enough autonomy slows down value realization. The goal is bounded autonomy-agents operating at speed with clearly enforced policies, escalation routes and explainability built in.
Strategic partnerships are becoming part of the underlying architecture. No organization can manage orchestration, integration, governance and operating model redesign alone at the speed now required. The most effective partnerships are co-engineered around shared accountability for outcomes, resilience and speed to value. Organizations also need to make time for alignment, bringing architects, engineers and risk leaders together early through design cycles and cross-functional roadmapping.
Why this matters for operations professionals
The organizations setting the pace understand that value at scale depends on becoming structurally ready to run AI-native operations. For operations teams, this means leading the redesign of governance, architecture and operating models-not waiting for AI to fit into existing structures. The gap between experimentation and production will close fastest where teams treat AI for Operations as a discipline that combines runtime control, clear accountability and bounded autonomy from day one.
Your membership also unlocks:
