Rising costs, uncertainty on the business value derived, and inadequate controls may be casting a shadow of doubt on the real power of Agentic AI behind all the hype.
Over 40 per cent of Agentic AI projects will be cancelled by the end of 2027, recent research by global technology consulting firm Gartner notes. Gartner, in its research, goes on to say that vendors too are contributing to the hype by engaging in “agent washing,” which refers to a rebranding of their existing products such as AI assistants and chatbots.
Gartner estimates that only about 130 of the thousands of agentic AI vendors are real. It stresses that many of the use cases positioned as agentic today don’t require agentic implementations at all.
“Most agentic AI projects right now are early-stage experiments or proofs of concept that are mostly driven by hype and are often misapplied,” Anushree Verma, Senior Director Analyst, Gartner, said. “This can blind organisations to the real cost and complexity of deploying AI agents at scale, stalling projects from moving into production,” Verma added, urging organisations to cut through the hype to make decisions about where to apply agentic capabilities.
However, the research firm predicts that at least 15 per cent of day-to-day work decisions will be made autonomously through agentic AI by 2028, up from 0 per cent in 2024, counting it as a leap forward in AI.
Speaking about Agentic AI, Ramprakash Ramamoorthy, Director of AI, Zoho, said that they are already seeing “serious exploration” from financial services, IT, and logistics sectors given their digital maturity. “That said, not every business problem needs an agent...The key is to separate hype from value, and that takes a technology partner who is playing the long game,” he adds. He believes that Indian enterprises tend to leap frog technologies, and this will make it easier for them to adopt agentic systems compared to companies in developed economies.
Soumendra Mohanty, Chief Strategy Officer of data science solutions firm Tredence, said that while large-scale deployment is still nascent, “the intent and investment” are real, and both experimentation and adoption are picking up. Today, approximately 10–15 per cent of Tredence’s client implementations have progressed into production, particularly in areas such as marketing automation, supply chain optimisation, sales operations, and others, he adds. Indian companies are still in the early stages as the technical infrastructure and AI readiness in India too are still evolving, he adds.