I think this is actually a place that a well implemented AI would potentially help "decent" call center employees. I remember working in a MCI call center years ago, like in 2003. The computers were ancient, you couldn't have more than one website open at a time (no tabs or multiple browser windows allowed) each click took 10 seconds or more to load, AND you had to go in and out across 3-5 systems. So while I knew were stuff was, if I hadn't memorized it by accident, or in the 1 week training, I would have to ask people to wait minutes as I switched around, and this assumed I remembered the exact place to go to do the lookup and didn't get lost.
However, I also remember I was the only one (having gone to college before) who wasn't massively surprised by the density and speed of content for the training week of the hires. And we were in upstate NY so no language barrier. They're not paying for educated or even quick thinking, they're not investing in the documentation or technology or training to make people good at all the weird edge cases (and especially today, about the only reason 90% of people call in is hitting an edge case). They're also not
@dioxide45 - these call center people likely have never used the system, or have access to the owner site to know about the points calculator etc.
The biggest issue is just that the KPI is time on a call. Lower is better. They're clearly not getting graded on quality of the answers or service. So they probably just work to the metric. I fully blame the management for all of this - they don't want to spend the time or money to improve the system.
On the AI side I finally had a success today with Claude Sonnet pulling together lots of information around a open source product integration between Foreman and Puppet. I was able to send it log snippets and it found relevant bug reports and forum threads to synthesize the issue, as well as give various useful options to parse existing logs to find more of the root cause as well as improve logging for the future. This is likely at or above the data synthesization level needed to go over all the odd rules and interactions in MVC points and explain as well as point the user to the info themselves so they can reference. Of course, it also cost my work ~$0.70 which is quite a lot as far as API AI goes for text and web parsing. I doubt MVC want's to pay that amount per interaction (figure time and number of queries and follow ups I'd guesstimate it might be over $5/hr if run continuously) as I do wonder if that wouldn't end up more expensive than the current (likely) offshored phone reps. Then again, that would probably need to be in addition to the current reps, at least for a while as I haven't been impressed with many of the current AI CS replacements I've encountered.
because I had to do 2FA to be able to login. What a PIA. I know some people think it makes things more secure, and it does, but it is a nuisance. Is someone really going to log in to my account and book my weeks?