Joey Levin
Analyst · Wells Fargo. Please go ahead
Yes. I'll do the Dotdash one first, and then I'll come to the AI one. The -- we're not so focused on audience extension there. We're focused on proving and showing intent in the product that we have. It has the key features that advertisers are looking for today, which is anonymity or you don't need PII, you don't need cookies and you have one of the hardest things to come across, which is intent, while not having those other two things. And the fact that we can deliver that, we're branding that, we're organizing around that. I think it's very helpful towards sales. But the focus on that is internally. Obviously, there is -- there are opportunities for audience expansion and there are components where that can happen. But I think that we have a lot of inventory that we can sell across our sites, and that's going to be the -- your priority and focus of that today, but not to say that those things aren't possible. And that comes out, I think, in the next few weeks, and so that will be exciting. We'll see how that goes. In terms of using AI in the businesses, there's a bunch of places where we're using -- probably where it's happening fastest and most robustly right now is in code. Meaning people writing code and using pieces of code that they can get through these systems or helping them right go faster that -- if you talk to pretty much any developer in any company, they are using it and they're getting real value from that really quickly, which leads to efficiencies there. There are other areas like customer support, where we have ideas or we're experimenting, but they haven't quite gone live with those things yet. And then just organizing processes around content creation. We're certainly not having the AI radar content, but you can start to organize and outline things and figure out how to prioritize things or use AI to learn what kind of content works, what kind of content works better than other content and analyze data, which is a data analysis project at significant scale, and that's working -- starting to work for us. When I think about -- I alluded to this in the letter, the marketplace business is, this is, I think, maybe one of the most exciting things, although we have nothing live here yet. But one of the most exciting things is to use these models to learn our proprietary supply database and learn from our demand, the demand that's coming in and figure out how to make better matches. Anytime you figure out how to make better matches that has significant yield for the business and making better matches is a data analysis problem. And these models are built for big scale data analysis. And I think that that will be -- could be very valuable. The other thing that I alluded to in the letter, I think is really important is take Angi, for example, we have what we call a service request path. On the service request path for Angi, people come in, we ask them a question, their zip code, then we ask them what kind of job it is, then we ask him some details about that job, and that is a multi-step process. It is hard to get a user anywhere through a multi-step process. What the chat bots are doing right now is they're creating this natural conversational UI where users are getting comfortable with those things, which is like a gift from heaven for us to be able to get people to use that UI more comfortably and to have that be smarter and more interactive. So we've actually built on already at Angi. It's not live yet, but we've built that. We're putting more work into building that to get something really exciting there, and that will be fun. But using that conversational UI to get better data from the homeowner on what they need done, and then, therefore, match them better on the service professional side. And you can make the same thing on the care side and on Vivian and Turo that same thing works. And I think that that could be a lot of fun and really impactful for the business.