Survey finds slow adoption of autonomous AI agents in enterprises

A survey by Gartner has shown that only 15% of IT application leaders are considering, piloting, or deploying fully autonomous AI agents within their organisations.

The research, which gathered insights from 360 IT application leaders in North America, Europe, and the Asia/Pacific region, found that the majority are engaged with AI agents to some degree, but fully autonomous implementations remain limited. The survey focused on the deployment, trust, and perceived impact of generative AI and agentic AI across enterprise applications.

Adoption and barriers

The report noted that while 75% of respondents are either piloting, deploying, or have already deployed AI agents in some form, concerns about governance, solution maturity, and agent sprawl are slowing the move toward fully agentic AI. These agents, defined as goal-driven AI tools requiring no human oversight, often face pushback due to security and organisational readiness issues.

Trust remains a significant barrier. Only 19% of respondents reported high or complete trust in their vendor's ability to provide adequate hallucination protection. Hallucination protection refers to mechanisms preventing AI systems from generating incorrect or misleading outputs. Additionally, 74% said they consider AI agents to be a new attack vector, and only 13% strongly agreed that their organisation had effective governance frameworks in place to manage the new risks associated with agentic AI.

Gartner's Senior Director Analyst, Max Goss, explained the situation:

"The hype around agentic AI continues to grow, with vendors positioning AI agents as the next phase of AI evolution that will address the shortfalls of more traditional GenAI assistants. Seventy-five percent of survey respondents said they were piloting, were deploying or had already deployed some form of AI agents into their organisation; however, concerns around governance, maturity and agent sprawl continue to hamper the deployment of truly agentic AI."

Impact assessment

The findings indicate a tempered expectation regarding transformative change driven by AI agents. Only 26% of respondents felt that agents would have a transformative impact on productivity. The majority, 53%, said the impact would be significant but might not be transformative, while 20% believed the gains would be marginal.

Alignment across the organisation appears to play a role in perceptions of AI's impact. Just 14% strongly agreed that there was clarity between IT, business users, and leadership on what problems AI should solve. Those with greater alignment were 1.6 times more likely to view AI agents as transformative, and more than three times as likely to extract significant value from their generative AI tools.

Goss emphasised the importance of this alignment:

"Alignment between IT, the business and executive leadership over what problems AI can solve and how to measure its value are critical for successful AI deployments, but we see that many organisations do not have this."

Deployment focus

Many organisations may not be focusing their AI deployments in the most effective areas. Those with less business alignment are nearly twice as likely to see office productivity as the top domain for AI agents, whereas organisations with clearer goals tend to prioritise vertical use cases such as customer service, ERP, and sales.

As Goss stated,

"Office productivity and the digital workplace are the default for those organisations that don't have a strong grasp on what they are doing with agents, but they are not necessarily the areas that will provide organisations with the most value. Among all survey respondents, analytics and business intelligence topped the list, with 64% ranking it among the top three domains that will be most impacted by AI agents, followed by customer service (55%) and office productivity (39%)."

Replacement of applications and workforce

The replacement of applications and workers by AI agents within the next two to four years remains unlikely, according to most leaders. Only 12% strongly agreed that agents would replace applications, and just 7% strongly believed they would replace workers within that timeframe. However, when considering those who somewhat agreed, the percentages rose: 34% believe applications will be replaced and 29% think workers will be replaced by AI agents in the coming years.

Goss commented,

"This is still significant for technology that has only been generally available for the last 12 months. It points to both the hype and the fear that exists around AI, specifically agentic AI."

Recommendations for organisations

Gartner analysts advise organisations to concentrate on three key areas for the deployment of agentic AI:

• AI agent governance: Build a platform-agnostic governance framework to reduce sprawl and set clear policies for safe use of AI agents across different tools and domains.
• Target AI agents at high-impact domains: Achieving alignment between IT and business teams on the specific problems AI should address will increase the likelihood of realising value from these technologies. If risks and returns in office productivity scenarios are uncertain, organisations may find greater benefits in applications for customer service or analytics.
• Adopt a multivendor strategy: Given the current immaturity of the technology, organisations should not rely on a single vendor for AI agent solutions and instead assess multiple options across ERP, CRM, and digital workplace portfolios.

The Gartner report concludes that the adoption of agentic AI remains in its early stages, with attention now focused on developing governance structures and identifying high-value use cases across enterprises.