A new Gartner analysis warns that over 40% of current agentic AI initiatives are likely to be abandoned by the end of 2027 due to rising costs, unclear business value, and a lack of sufficient risk controls. While agentic AI – artificial intelligence that can independently make decisions and take action to accomplish goals – is widely viewed as the next frontier in enterprise automation, the current wave of early experimentation may be moving too fast for its own good.
Anushree Verma, Senior Director Analyst at Gartner, explained that much of the current activity in the space is being driven by hype rather than viable implementation strategies. “The majority of agentic AI projects at the moment are proof-of-concepts or early-stage experiments. They are frequently misapplied, which can hinder their path to production. Enterprises often underestimate the complexity and cost of deploying AI agents at scale.”
A Gartner survey conducted in January 2025 among over 3,400 webinar participants revealed mixed levels of commitment to agentic AI. While 19% of organizations reported making significant investments, 42% were investing cautiously. Another 8% had made no investment at all, and 31% indicated a hesitant or observational approach. This broad spread reflects the uncertainty many organizations feel about the practical business utility of agentic AI today.
Adding to the confusion is the phenomenon Gartner calls “agent washing” – the rebranding of existing technologies such as chatbots, digital assistants, or robotic process automation (RPA) as agentic AI, even when they lack the core capabilities. Of the hundreds of vendors currently touting agentic AI solutions, Gartner estimates that only around 130 offer authentic agentic functionality.
Agentic Implementations
According to Anushree Verma, most available models lack the autonomy to deliver long-term commercial value. “A lot of use cases that are framed as agentic don’t actually require agentic implementations. The return on investment simply isn’t there yet in most scenarios,” she said.
Despite these early hurdles, Gartner notes that agentic AI holds real long-term promise. By 2028, the firm predicts that 15% of daily workplace decisions will be made by agentic AI, up from virtually zero today. Moreover, agentic AI is expected to be embedded in 33% of enterprise software applications within the same timeframe, compared to less than 1% in 2024.
However, Gartner urges companies to avoid rushing into deployments unless the business case is strong and clearly articulated. Integrating AI agents into legacy systems can be complex and expensive, often disrupting workflows and requiring deep architectural changes. In many cases, organizations may need to redesign processes from the ground up to fully leverage the technology.
“Companies should focus on enhancing overall enterprise productivity, not just automating individual tasks,” said Anushree Verma. “Start small – use AI assistants for simple data retrieval, automation tools for routine tasks, and reserve agentic AI for scenarios that involve complex decision-making. It’s about driving business value through a balance of cost, quality, speed, and scale.”
While the promise of agentic AI remains compelling, Gartner’s findings suggest a more measured, strategic approach is necessary to separate potential breakthroughs from unsustainable hype.
