Enterprise AI readiness trails the hype amid agentic rush

Enterprise AI is moving, but not nearly as fast as the rhetoric around it. That was the blunt message from David Linthicum, founder and lead researcher at Linthicum Research, who said many companies are still working out infrastructure and cost before they can push AI into real business operations.

Right now, most adoption is still centered on large language models, edge systems, and narrow uses like calendaring, software development, and process automation. What hasn’t happened yet is the broader shift into supply chains, inventory management, and other core functions where the data, security, and governance bar gets a lot higher.

Linthicum said that at VMware Explore, the companies he spoke with were not ready for AI systems yet. They were looking for a partner to modernize infrastructure and get them to the point where AI can run on premises. Broadcom’s VMware AI Factory, built on VMware Cloud Foundation, is aiming at exactly that problem by automating the path from bare metal to model deployment.

But the bigger warning was about agents. Linthicum argued that most agentic projects add complexity, overhead, and security risk without actually needing an agent architecture. His rough estimate was that 95% of the agentic applications he sees do not need to be agentic at all. That is a sharp way of saying the market may be reaching for autonomous software long before it has a good reason to use it.