[kjs3]

Even more, when the concept of "AI agent" appeared on their radars, they were the first people who they thought they're gonna replace.

I worked on a project where they replaced virtually all tier-1 customer service reps with an AI chatbot, which worked ok, because 90% of those calls are things like "what's my balance" and "I need a replacement credit card", which an AI can competently navigate. Phase 2, I was informed, was to replace Tier-2 with a chatbot, so the only time a human has to get involved was the most complicated problems. SO MUCH MONEY SAVED AMIRIGHT?!?

So I asked, in the middle of this giant kickoff meeting, "if you get rid of tier 1 & 2 reps, how are you going to train up the tier 3 reps who know enough about how things work to handle these most-complicated problems?". After an extended silence we got "we have a plan for that but we can't go into it just now". Sure you do.

[sceptic123]

The plan is to replace the tier 3 reps too

[vanuatu]

Do you need to do tier 1 and 2 work before tier 3?

If they are structurally different, and there’s a way to train people directly into tier 3, then it doesn’t seem unreasonable to automate t1 and t2 as from my experience the vast majority of the tickets are either simple or repeated workflows. Taking the idea to the limit, you’d automate all tiers, and have the ai escalate to the individual teams within the company for any truly meaningful edge cases

I feel sort of the same about SWE, which is much more complex, but juniors can ostensibly grow into seniors with AI

[kjs3]

If it wasn't clear, I'm talking about CSRs who deal with people, not computers.

[stronglikedan]

> how are you going to train up the tier 3 reps

iterative feedback loops using memories for context, of course. just like they trained the first two tiers

[kjs3]

Where are the feedback loops for customer issues that haven't arisen before, or happen very infrequently. The existence of a tier 3 is for problems that aren't an everyday occurrence. Further, these are often all or partially 'people problems', not 'technology problems'.