Gartner: Over 40 per cent of agentic AI projects will be discontinued by the end of 2027
Reasons include rising costs, unclear economic benefits, and insufficient risk management measures.
“Most of these projects are still in an early phase – they are often experiments or proof-of-concept initiatives that have emerged from the current hype and are frequently misapplied,” explains Anushree Verma, Senior Director Analyst at Gartner. “This often leads companies to underestimate the real costs and complexity involved in scaling AI agents – as a result, many of these projects never reach productive operation. It is crucial to see through the hype and make strategically sound decisions about where and how this new technology can truly be used to add value.”
A Gartner survey of 3,412 participants in a webinar in January 2025 revealed that 19 per cent of respondents stated their organisation had already made significant investments in agentic AI. 42 per cent reported more cautious investments, while 8 per cent had not invested at all. The remaining third were either undecided or taking a wait-and-see approach.
Agent Washing
Another driver of the hype: many vendors engage in so-called “agent washing” – renaming existing technologies such as AI assistants, robotic process automation (RPA), or chatbots as supposedly agent-based solutions, even though these often lack the essential characteristics of genuine agents. Gartner estimates that out of thousands of vendors, only around 130 actually offer authentic agent-based AI technologies.
“Most of the currently available offerings lack substantial benefits or a genuine return on investment (ROI),” Verma emphasises. “The underlying models are still technically immature. They lack the necessary autonomy to independently achieve complex business objectives and are unable to reliably follow nuanced instructions over an extended period. Many use cases marketed today as agent-based do not, in fact, require agent-based implementations.”
Agentic AI with Potential
Despite these early-stage challenges, Gartner believes agentic AI represents a significant step forward in AI capabilities and opens up new market opportunities. It offers the potential to use resources more efficiently, automate complex tasks, and drive innovation in everyday business processes – far beyond what conventional script-based automation bots or virtual assistants can achieve today.
Gartner forecasts that by 2028, at least 15 per cent of all daily business decisions will be made autonomously by agentic AI. Furthermore, by 2028, around one-third of all enterprise software applications are expected to include agentic AI components.
Added Value and ROI Must Be Evident
In this early phase, Gartner recommends deploying agentic AI only where it provides clear added value or demonstrable ROI. Integration into existing (often legacy) systems is frequently technically complex, can significantly disrupt established processes, and often requires costly adaptations. In many cases, a fundamental redesign of workflows is the better strategy for successful implementation.
“To unlock the true benefits of agentic AI, organisations should not focus solely on automating individual tasks but place enterprise-wide productivity at the centre,” says Verma. “A good starting point is the use of AI agents for decision-making situations, automating routine processes, or handling simple requests. The goal is to maximise business value – whether through lower costs, better quality, increased speed, or improved scalability.”
