The appetite is enormous. Deloitte’s early-2026 research found that close to three-quarters of companies plan to deploy agentic AI within two years. Yet only 21% of those companies say they have a mature model for governing what those agents are allowed to do. That gap matters: an agent with system access can approve the wrong payment as quickly as the right one. Deloitte titled its companion analysis bluntly: agentic AI is “scaling faster than guardrails”.

Not everyone is convinced the substance matches the noise. Gartner predicts that more than 40% of agentic AI projects will be cancelled by the end of 2027, “due to escalating costs, unclear business value or inadequate risk controls”. Its analyst Anushree Verma is blunter still, calling most current projects “early-stage experiments or proof of concepts that are mostly driven by hype and are often misapplied”. Gartner even has a name for the marketing sleight of hand, “agent washing”, the rebranding of old chatbots and automation as agents. Of the thousands of vendors now claiming agentic capability, it reckons only around 130 are the real thing.

None of this means the direction is wrong. When Anthropic released Claude Sonnet 5 in June 2026, it described a model that could “make plans, use tools like browsers and terminals, and run autonomously at a level that, just a few months ago, required larger and more expensive models”. The capability is arriving, and it is getting cheaper by the month.

For a sector built on oversight and accountability, this is familiar ground. The organisations that get value from agents will be those that let their agents earn autonomy slowly, under close supervision. The question is no longer whether AI can act on its own, but how much we should let it, and under what controls. Industries that already live under strict oversight, drug safety among them, may be better prepared for it than most.

“Most businesses do not need an autonomous agent tomorrow. They need one repetitive task done reliably, with a human still in the loop. That is where the value actually starts.” Amit Jaitly, Director of JMC Solutions