AI Agents Aren’t Ready for Consumer-Facing Work—But They Can Excel at Internal Processes

Companies can create value with agentic AI, but they often start in the wrong place. While it can be tempting to use agentic AI for customer-facing applications, those tend to be complex, messy, and unforgiving of errors—not ideal for agentic AI. Back-end operations, however, are a better fit. A project at a European telecom illustrates what this looks like. Focusing on reducing the time and cost for service calls, the company was able to use agents to reduce resolution time by 60%, save more than a million euros on a recurring annual basis, and significantly improve the company’s net promoter score. This example offers lessons for leaders, but also illuminates the common challenges they’ll need to navigate in this process.