Why are companies deploying agents faster than governance matures?

Deloitte's survey of enterprise AI adoption finds that "Agentic AI is poised for growth with close to three-quarters of companies planning to deploy Agentic AI within two years," but that "only 21% of those companies report having a mature model for agent governance." The gap sits inside a broader picture of rapid scaling: workforce access to sanctioned AI tools grew "from fewer than 40% to around 60% of workers" in a year, and 85% of companies expect to customize agents "to fit the unique needs of their business." Deloitte frames the enterprise as racing to deploy agents faster than it is building the oversight to run them safely.

Deloitte's own account of what separates success from failure is not more automation but a sequencing discipline: "Companies seeing the most success are taking a measured approach—starting with lower-risk use cases, building governance capabilities and scaling deliberately." The report's framing line makes the causal claim explicit: "governance is more than guardrails—it's the catalyst for responsible growth," i.e., Deloitte argues governance is a precondition for scaling value, not a brake on it.

This sits in tension with Why do production AI agents stay deliberately simple?, which found practitioners actually building agents today constrain autonomy themselves, bottom-up, to keep systems reliable. Deloitte's figures describe the opposite posture at the leadership level: intent to scale agentic deployment broadly over the next two years while governance capability lags behind that intent. The gap also compounds what What collaboration level do workers actually want with AI? describes as misalignment between worker-preferred autonomy levels and actual deployment choices — immature governance is one likely reason that misalignment persists. It also bears on Where have workers actually delegated tasks to AI?, since the information-intensive work where delegated AI already concentrates is exactly where agentic governance gaps would be most consequential.

The excerpt does not say what counts as a "mature model for agent governance," so the 21% figure cannot be checked against a defined standard, and it gives no sample size or methodology for the underlying survey — these are Deloitte's self-reported figures from companies it surveyed, not independently verified outcomes. Deloitte is also a consultancy that sells AI strategy and implementation services, which gives it a commercial interest in both documenting urgency around agentic AI and prescribing the "measured approach" of governance-building it advises; the finding is suggestive of a strategy-versus-execution gap in the market Deloitte serves, not a demonstrated causal account of why agent governance lags deployment plans.