In my professional opinion, AI in support should be implemented in small, thoughtful, methodical bursts. Test, test, and test again. One policy at a time, reaching your own standard of quality before moving on. Don't overthink, but keep a close eye on outputs. Involve your frontline team so they have visibility into AI responses and feel welcome to give feedback for improvement. They are closest to the customer and will catch what the metrics miss. AI in support is there to handle the low-hanging fruit so your human team can focus on the deep, meaningful interactions that generate real brand love.
