A contact center used AI to cut the time agents spend idle between calls. The system listens to live calls, extracts key information, and surfaces what agents need to know. They later added a trend tracker showing that 70% of current callers ask about the same issue, so the whole team can prepare.
Chapters
So this contact center, we had to then find what's the one metric we can use to measure the success of those helpers. And we didn't want to immediately be like, you're all sellers now. We instead decided to focus on the amount of time they spent not talking to customers. It's called non-talk time in the industry. And so how do we eliminate or basically completely remove non-talk time? So at the beginning of the process, we created an LLM-based system that would listen to the conversation, do entity extraction. And then surface up topics so that the person on the call could at least easily search for what they needed to find. Sure. But now we are like, well, wait, we could actually add an additional model that will index and search for content. Then we could add a global trend identifier. So it's looking at all the agents and finding out, wait, right now 70% of people are asking about this thing. Now we surface that to everyone.