“The top challenge with agentic frameworks is that each vendor takes a fundamentally different approach to agent architecture, state management, and communication protocols. As vendors push their own frameworks and agents, enterprises will face adoption challenges, including a significant rise in technical debt and maintenance overhead,” Liddle says. “I don’t see a single framework emerging to unify all agents, making this complexity an ongoing reality.”
Meanwhile, enterprise vendors who are rolling out agentic AI services as part of their flagship offerings make their pitch for a platform-based approach to managing agents.
“Managing agentic AI at scale is a multidimensional challenge. To break it down, it is a governance, operational, ethical, and integration challenge all at once,” says Chris Bedi, chief customer officer and enterprise AI advisor at ServiceNow. “To manage agentic AI, one needs a platform that can unite AI agents, data, and workflows, with a single data model, which brings AI to every corner of the enterprise and addresses all these challenges.”
