With current real-world deployments struggling to meet business expectations, market analysts warn most agentic AI initiatives may remain experimental beyond 2028.

Warnings about the future of agentic AI are reverberating in some regions, as new forecasts suggest that more than 40% of such projects could be abandoned by the end of 2027.

The prediction, released by technology advisory firm Gartner, is drawing attention not just for its scale but for the underlying reasons: escalating costs, unclear business value, and inadequate risk controls — that threaten to derail the next wave of AI innovation.

However, a closer look at Gartner’s forecast reveals that these are not certainties grounded in broad, industry-wide data. Rather, the firm’s projection is based on a combination of its own market analysis and a January 2025 poll of 3,412 webinar attendees. In that poll, 19% of respondents had indicated plans to make significant investments in agentic AI, while 42% had cited a more cautious approach. Some 31% were either undecided or waiting to see how the technology evolves. These figures suggest that, despite the hype, most respondents remain tentative, with a minority committing substantial resources to agentic AI at this stage.

According to the firm’s Senior Director Analyst, Anushree Verma: “Most agentic AI projects right now are early stage experiments or ‘proof-of-concept’s that are mostly driven by hype and are often misapplied.” The firm cautions that organizations risk underestimating the complexity and cost of deploying AI agents at scale, which could ultimately lead to a wave of cancellations and failed initiatives.

Beware of predictions and hype
Complicating the picture is “agent washing”: the rebranding of existing AI assistants, robotic process automation tools, and chatbots as agentic AI, even when they lack genuine autonomous capabilities. Only about vendors out of thousands truly offer agentic AI solutions, according to the firm.

Elsewhere, reports are noting that real-world deployments of agentic AI have frequently struggled to deliver on their promises. Technical immaturity, integration challenges, and a lack of robust risk controls have contributed to a high rate of project failures. In practice, many organizations have found that agentic AI systems are not yet capable of autonomously achieving nuanced business goals, or reliably following complex instructions over time.

Based on limited data, the firm predicts that by 2028, agentic systems could autonomously handle at least 15% of day-to-day work decisions and be embedded in a third of enterprise software applications. What is more certain is that industry experts and analysts alike would urge organizations to approach agentic AI with caution, focusing on clear business needs, realistic outcomes, and strategic fit before making significant investments.