Gartner research has found that most organisations have yet to adopt AI agents at scale, despite growing industry attention. In a survey of 360 IT application leaders conducted in May and June 2025 across North America, Europe and Asia/Pacific, only 75% of respondents said they were piloting, deploying or had already deployed AI agents. Analysts described the result as modest given the level of market hype.
The survey highlighted concerns that are slowing adoption. Just 19% of respondents said they had high or complete trust in vendors’ ability to provide adequate protection against hallucinations, while 74% viewed AI agents as a potential new attack vector. Only 13% strongly agreed their organisations had the right governance structures in place to manage agentic AI. Gartner analysts pointed to trust, security, governance and organisational readiness as the principal barriers to more widespread deployment.
Expectations of productivity gains from AI agents were mixed. While 26% of respondents said they would have a transformative effect, a majority of 53% believed the impact would be significant but not transformative, and 20% expected only marginal improvements. The research found that organisations with strong alignment between IT, business users and executive leadership were substantially more likely to report higher levels of value from AI initiatives.
The study also indicated that many organisations may be targeting AI agents in areas with limited returns. Firms without a common understanding of AI’s role were nearly twice as likely to prioritise office productivity tools. By contrast, those with clearer alignment directed efforts toward vertical functions such as customer service, ERP and sales. Across all respondents, analytics and business intelligence was the domain most frequently cited for AI agent impact (64%), followed by customer service (55%) and office productivity (39%).
Looking ahead, most leaders did not expect AI agents to replace applications or jobs within the next two to four years. Only 12% strongly agreed they would replace applications in that period, and 7% strongly agreed they would replace workers.
