Empowering teams to unlock insights faster at OpenAI

I truly believe that great products are built from a place of customer centricity. User Ops gets millions of tickets every year from our customers and we realized that we were sitting on this treasure trove of information. I'm fortunate to work with a lot of different stakeholders who are deeply curious about our customers. They want to understand what do they like about our products? What do they not Where are they experiencing friction? inspiration? But in order to do that, it's so manual. It required a ton of prioritization and it was cutting off our curiosity. So, we wanted to be able to scale how we did that work, which is how this research assistant was born. I think about this tool from a place of being able to choose your own adventure. Some people would come in and they'd want to be able to quickly visualize what is the volume of our customer support tickets? How has that volume changed over time? What are the major themes in a pretty structured way around those tickets? So, there's one front end that allows you to get that visual overview. And there is another part of the tool that allows you to go in with a question and talk to the data the way that you would to chatt. So you can ask a very specific question about our customers, about a product that they're using, about a policy that we've rolled out, and you can see how customers are responding to that very thing. With SQL or another dashboarding tool, you have to have a very specific idea of what you want to look for. So the tools are extremely static. The difference here is that we can define the questions in real time or any user that they care about. The constraint is the data set which is extremely flexible. It's not the questions that you want to answer. So now in order to access insights about customers, you don't need to have a background in coding or SQL or advanced analytics. All you need to do is come in with a lot of curiosity and some good questions about our customers.