LinkedIn·Thursday, 11 June 2026·11 Jun 2026

We let AI handle a customers Tier 1 support tickets for 90 days. Here's the number that surprised us and the one that scared us. The…

Airobi

175 followers

We let AI handle a customers Tier 1 support tickets for 90 days. Here's the number that surprised us and the one that scared us. The surprising one: 68% of Tier 1 tickets resolved with zero human involvement. Going in, our team bet on 40%. We were prepared to walk it back if customers pushed away from it. They didn't. CSAT on AI-resolved tickets came in at 4.6/5 — slightly higher than our human baseline. Faster resolution turned out to matter more to customers than who, or what, was on the other end. That's the part everyone wants to hear. Lower cost per ticket, faster response times, agents freed up for complex work. The business case wrote itself. Then we looked at the second number. Of the 32% that escalated to humans, 71% were the same handful of issues recurring problems the AI couldn't solve because they weren't support problems at all. They were product gaps, billing errors, and broken internal processes. For years, those issues had been quietly absorbed by support agents. Smart, patient people working around the cracks so customers never felt them. The automation didn't create those cracks. It just removed the cushion that had been hiding them. That's the number that scared us: not what AI couldn't handle, but how much dysfunction the support team had been silently subsidizing. The lesson wasn't "automate Tier 1." It was that your support queue is the most honest product roadmap in your company. AI just makes it impossible to ignore. We didn't need fewer agents. We needed to stop using them as a patch. What's the most common ticket your team has quietly normalized?