The real cost, security, and culture problems behind enterprise AI agents

At VentureBeat's AI Impact event, Red Hat's Brian Gracely argued that many enterprises are not as far behind on AI agents as they fear, but production deployments quickly expose hard problems. He said agentic systems drive much higher usage than earlier chatbots, making cost discipline and dependence on a few model providers major concerns.

The discussion signals a shift from AI agent experimentation to operational governance, where infrastructure choices and model sizing can determine whether deployments scale economically. Enterprises may increasingly pursue provider diversification and tighter controls to reduce vendor lock-in, security exposure, and runaway spending.