More than 40% of projects involving agentic artificial intelligence—software agents capable of independently executing tasks—will be cancelled by the end of 2027, Gartner predicts in a report released Wednesday. The research firm cites escalating costs, unclear business value and inadequate risk and security controls as the chief reasons many initiatives will fail to make it into production. Gartner warns that the market is rife with “agent washing,” estimating that only about 130 of the thousands of vendors promoting agentic offerings provide products that genuinely meet the definition. The finding comes even as enterprise software makers such as Salesforce and Oracle pour billions of dollars into AI agents in hopes of improving margins and automating complex workflows. Despite the anticipated shake-out, the consultancy expects continued uptake of the technology. It forecasts that agentic systems will account for at least 15% of day-to-day work decisions and appear in 33% of enterprise software applications by 2028, up from virtually none today. A January survey of 3,412 executives showed 19% of organisations have made significant investments in agentic AI, while 31% remain on the sidelines. “Most agentic AI projects are early-stage experiments driven by hype and often misapplied,” said Anushree Verma, a senior director analyst at Gartner.