Briefing

  • A report released by Dataiku on Thursday shows that nearly three-quarters (about 74%) of CIOs regret major AI vendor or platform decisions made in the past 18 months, with 62% of respondents saying they have been directly questioned or challenged by their CEO over vendor choices. The report was compiled by Harris Poll on behalf of Dataiku after surveying 600 CIOs globally.
  • The deadline to prove the value of AI investments is approaching: 71% of CIOs say AI budgets are likely to be cut or frozen if goals are not met by mid-2026. Meanwhile, 85% of IT leaders say gaps in traceability and explainability have caused AI projects to be delayed or prevented from reaching production.
  • "CIOs are moving from experimentation to accountability faster than most organizations anticipated," said Florian Douetteau, co-founder and CEO of Dataiku, in a press release.

Deep Insights

As CIOs face critical scrutiny over AI vendor choices, they also need to address how to manage employee use of AI tools.

Among the CIOs surveyed in the Dataiku report, more than half said they had discovered employees using unauthorized shadow AI tools, applications, and platforms to complete work tasks. 89% of CIOs believe that unconstrained AI access will create significant technical debt within their organizations.

Another report released in January by security company BlackFog shows that 6 out of 10 enterprise employees say they would use shadow AI tools if they helped them meet work deadlines, and 86% of employees use AI tools at least once a week to complete tasks.

The Dataiku report also reveals that employees are creating AI agents and applications faster than IT departments can govern them, with 82% of CIOs believing this exacerbates the "agent sprawl" problem. The report notes that as enterprises embed AI agents into critical business processes, governance gaps are widening.

Agent sprawl is a common pain point for IT leaders across industries, with companies in retail, food and beverage, and other sectors deploying the technology to improve efficiency and revenue. However, research firm IDC predicts that by 2029, the number of active AI agents globally will exceed 1 billion.

The Dataiku report found that while 87% of CIOs say AI agents are embedded in critical systems, only 25% say they have full visibility into all agents in production environments.

"The pressure is real," said Douetteau. CIOs need to build AI systems that can "explain, govern, and be accountable for, before accountability is imposed rather than chosen."