In an analysis published by Shep Hyken at Forbes, United Airlines is fundamentally restructuring its contact center operations to counter the unintended consequences of artificial intelligence. The carrier is moving away from rigid policy scripts in favor of empowering human agents with autonomous decision-making authority.
Speaking at Customer Contact Week in Las Vegas, Bryan Stoller, United Airlines’ VP and Global Head of Customer Care and Contact Centers, outlined a transition toward what he terms the “Reasonable Department.” The initiative is designed to address a growing industry paradox where automated efficiency inadvertently concentrates high-stress, complex escalations onto human workers.
Standard automated systems now handle routine inquiries, leaving representatives to manage only unique disruptions that lack a Standard Operating Procedure. To adapt, the airline has deployed specialized internal teams known as “hard stuff hunters” to identify and proactively resolve high-effort escalations before they reach the contact center floor.
This operational pivot operates on a defined four-pillar service model: Detect, Route, Resolve, and Learn. By implementing these pillars, United aims to provide its agents with the necessary frameworks to make independent, context-driven decisions when standard policies fail to address a passenger’s situation.
How Automated Deflection Amplifies Human Workloads
The aviation sector’s heavy investment in AI chatbots has successfully reduced call volume for basic tasks like flight status checks and baggage fee inquiries. This filtering process removes the straightforward interactions that previously offered human agents brief mental resets between difficult calls.
The remaining queue now consists entirely of complex, emotionally charged disruptions. By officially authorizing contact center staff to operate outside of black-and-white policy constraints, airlines are attempting to protect brand loyalty while mitigating worker burnout in an increasingly automated support environment.