81% of CIOs Lack Full Oversight of AI Agents (discover more) Built Outside IT, Dataiku Says
A Dataiku survey of 685 CIOs reveals significant gaps in AI agent oversight:
- 81% lack a full view of AI agents created outside formal channels.
- 72% cannot consistently confirm if these agents deliver intended business outcomes.
Key Findings:
- 90% of CIOs track all their agents, but only 72% can verify their performance.
- 67% estimate over 51 AI agents are currently in production.
- 83% lack standardized lifecycle management for agents across the organization.
- 47% have already decommissioned more than 20 agents this year.
- Only 21% have near-real-time visibility into AI costs by team or use case.
Speed vs. Governance:
84% of CIOs agree that employees are building agents and applications faster than IT can govern them, with 60% lacking a central AI governance layer.
Responsibility for Wrongful Agents:
Responses varied on who owns an agent that goes wrong:
- Shared teams (23%)
- Central IT (21%)
- Data or AI team (20%)
- Security, risk, and compliance (18%)
Career Implications:
88% of CIOs believe their success with AI will shape their reputation or career. 87% have been told by their CEO that job security depends on AI outcomes. 76% expect their role to be at risk if measurable gains from AI aren't shown by the end of 2027.
Budgetary Pressures:
72% expect a cut or freeze to their AI budget if performance targets are missed by the end of 2026. 97% report increased board pressure to demonstrate ROI on AI.
US CIO Concerns:
Dataiku found particularly stark numbers among US CIOs:
94% say employees build agents faster than IT can govern them. 87% expect budget cuts or freezes if targets slip. 82% regret at least one major AI vendor or platform choice from the past 18 months.
Looking Ahead:
75% plan to use more or different models, and 74% are considering open-source or open-weight models as a hedge.
Dataiku's Agent Management software aims to address these challenges by centralizing agent inventory and integrating with major cloud platforms. It goes on general sale in October, priced per instance annually, with monitoring charged per agent.
Gartner's Prediction:
In June 2025, Gartner predicted that over 40% of agentic AI projects would be cancelled by the end of 2027 due to rising costs, unclear business value, and inadequate risk controls. Amazon's AWS reports that almost 90% of early agent prototypes never reached production.