Katie Bianchi

The work we do is very technically complex and it's high stakes. So we tried to push AI into the deep end early, knowing that if it could help our teams handle the toughest cases, we would unlock real transformation.

The work we do is very technically complex and it's high stakes. So we tried to push AI into the deep end early, knowing that if it could help our teams handle the toughest cases, we would unlock real transformation.

The work we do is very technically complex and it's high stakes. So we tried to push AI into the deep end early, knowing that if it could help our teams handle the toughest cases, we would unlock real transformation.

But what it required us to do is really understand how every single case got resolved, which could be resolved simply through improving our documentation or improving our process so the AI could essentially learn from every interaction.

But what it required us to do is really understand how every single case got resolved, which could be resolved simply through improving our documentation or improving our process so the AI could essentially learn from every interaction.

But what it required us to do is really understand how every single case got resolved, which could be resolved simply through improving our documentation or improving our process so the AI could essentially learn from every interaction.

But then it also forced us by focusing on how complex cases get resolved as well to accelerate some of the decision points that we made around collecting better product telemetry, collecting more information from customers that may not fully resolve the case because they're very complicated, but would take the time to actually accurately diagnose that case from sometimes weeks to

But then it also forced us by focusing on how complex cases get resolved as well to accelerate some of the decision points that we made around collecting better product telemetry, collecting more information from customers that may not fully resolve the case because they're very complicated, but would take the time to actually accurately diagnose that case from sometimes weeks to

But then it also forced us by focusing on how complex cases get resolved as well to accelerate some of the decision points that we made around collecting better product telemetry, collecting more information from customers that may not fully resolve the case because they're very complicated, but would take the time to actually accurately diagnose that case from sometimes weeks to

down to days and hours. And by focusing on that full system and really understanding in the detail how the cases are resolved, with which data, with which expertise, it really helped us focus on everything to have a bigger impact at the start.

down to days and hours. And by focusing on that full system and really understanding in the detail how the cases are resolved, with which data, with which expertise, it really helped us focus on everything to have a bigger impact at the start.

down to days and hours. And by focusing on that full system and really understanding in the detail how the cases are resolved, with which data, with which expertise, it really helped us focus on everything to have a bigger impact at the start.

Absolutely. So, you know, we think about it in three steps, but I'll answer the last part of your question first. Like we committed to fixing the foundation. focusing on simple to complex issue resolution and how we built that infrastructure in alignment with our IT organization and with our product organization. And going back to first principles means that you have to be patient.

Absolutely. So, you know, we think about it in three steps, but I'll answer the last part of your question first. Like we committed to fixing the foundation. focusing on simple to complex issue resolution and how we built that infrastructure in alignment with our IT organization and with our product organization. And going back to first principles means that you have to be patient.

Absolutely. So, you know, we think about it in three steps, but I'll answer the last part of your question first. Like we committed to fixing the foundation. focusing on simple to complex issue resolution and how we built that infrastructure in alignment with our IT organization and with our product organization. And going back to first principles means that you have to be patient.

There's no shortcuts in innovation. And so that process for us around fixing the foundation was... all about understanding what data resolved our problems and how do we assemble teams to assess and quickly clean it up and load it and test again to make sure that the data that we loaded resolved the issue the second time. And then we had to completely rebuild our processes.

There's no shortcuts in innovation. And so that process for us around fixing the foundation was... all about understanding what data resolved our problems and how do we assemble teams to assess and quickly clean it up and load it and test again to make sure that the data that we loaded resolved the issue the second time. And then we had to completely rebuild our processes.

There's no shortcuts in innovation. And so that process for us around fixing the foundation was... all about understanding what data resolved our problems and how do we assemble teams to assess and quickly clean it up and load it and test again to make sure that the data that we loaded resolved the issue the second time. And then we had to completely rebuild our processes.

Our processes were built for case resolution pre AI, not post AI. So when you take the lens that you want to learn from every single interaction, you have to redo your processes. And so that was sort of the second step of that. And we knew if we didn't get that right, the AI wouldn't scale. So that's actually where I would say, you know, on a two plus, probably a two plus year journey,

Our processes were built for case resolution pre AI, not post AI. So when you take the lens that you want to learn from every single interaction, you have to redo your processes. And so that was sort of the second step of that. And we knew if we didn't get that right, the AI wouldn't scale. So that's actually where I would say, you know, on a two plus, probably a two plus year journey,