Hype vs reality: The rocky road ahead for agentic AI

A Gartner report warns that over 40% of agentic AI projects could be cancelled by 2027 due to rising costs, unclear value, and security gaps. While only a fraction of vendors offer substantial solutions, agentic AI still holds long-term promise. The report emphasizes the need to understand the roles and complexity of AI agents to unlock their true enterprise potential.

A new report by Gartner has revealed that over 40% of agentic AI projects are expected to be cancelled by the end of 2027, largely due to increasing costs, unclear business value, and inadequate security measures. The findings are based on a January 2025 survey involving 3,412 webinar attendees.

The report highlighted that 19% of organizations have already made significant investments in agentic AI, while 42% have opted for conservative investments. Another 8% have made no investments at all, and the remaining 31% are either undecided or adopting a wait-and-see approach.

“Most agentic AI projects right now are early-stage experiments or proofs of concept that are mostly driven by hype and are often misapplied,” said Anushree Verma, Senior Director Analyst, Gartner. “This can blind organizations to the real cost and complexity of deploying AI agents at scale, stalling projects from moving into production. They need to cut through the hype to make careful, strategic decisions about where and how they apply this emerging technology.”

The technology driving cognitive enterprises comes in the form of AI agents — autonomous systems that perform tasks depending on four key dimensions: the role they fulfill, the level of complexity they manage, their positioning within the enterprise, and the enabling technologies that power them. Understanding these dimensions is essential to grasp their potential and limitations.

In terms of roles, AI agents can inform by identifying patterns in data, predict future scenarios, execute tasks at scale, create content, recommend context-based actions, and even orchestrate multiple agents to address complex, interconnected challenges. Depending on how they are designed, agents can serve one or several of these roles simultaneously.

The complexity of AI agents ranges from basic to highly sophisticated. At the simplest level, they act as ultra-narrow tools designed to retrieve or answer a single, well-defined query with no reasoning involved. As they become more advanced, agents can autonomously coordinate workflows across departments, manage data harmonization, and handle intricate operational processes. At the frontier of development, agents are capable of supervising networks of specialized sub-agents, enabling adaptive, enterprise-wide decision-making and execution.

However, the Gartner report warns that current overhype — including what it terms “agent washing,” where companies rebrand traditional tools like AI assistants or RPA bots as agentic AI — can mislead buyers and stall innovation. Of the thousands of vendors active in this space, only about 130 were deemed to offer truly substantive solutions.

Despite these early-stage hurdles, Gartner maintains that agentic AI holds long-term promise. The report forecasts that by 2028, at least 15% of daily operational decisions in enterprises will be made autonomously by AI agents, up from 0% in 2024. Moreover, 33% of enterprise software applications are expected to integrate agentic AI by 2028, compared to less than 1% in 2024. These projections underscore the importance of moving beyond hype to build truly scalable, secure, and valuable AI solutions.