Yeah, I would probably first just say I don't know that those are binary, and I don't know that they're black and white. It's one or the other. But forced to choose, I think we're much closer to the former than the latter in the sense of the experiences that we power and that we're continuing to invest in being able to build out are always starting from a place of how do we best take advantage of our publishers content, which is written by humans. And then we think that's a really important ingredient here and leverage what's happening around AI and make it really easy for the publisher to maybe convert their content, maybe augment their content, but in one form or another enhance it, make the experience more interactive, more user -friendly, more personalized, more relevant through chat as the interface that we're building around. And so in that sense, content is the core of it. It's about the stories you've written, the things your journalists have researched, the articles you're publishing, and the topics that your readers are already on your site for. You already have this audience, and certainly a big part of what we do is getting you a new audience by distributing these chatbots in other platforms where maybe you didn't have a presence before because these are chat endemic environments like this one. Again, things are moving so quickly here, and the opportunities are so broad. There is sort of a spectrum, I think, between example A and then this companion assistant version of my AI or Chashi BT or what like the promise I think of Siri and Alexa was back in the day. And I do think things can move that direction even for content companies. And we've assisted for their content a librarian that's personalized to every reader. We've gotten those analogies in sort of private meetings from publishers that we're working with and talking to, and there's ways to take advantage of our platform to deliver that experience. I do think it's still different than what you would see from my AI on Snap where it's more or less, and not to diminish it, I think they've done tremendously well despite some of the challenges that you alluded to. It's more or less the Chashi BT experience, right? It's a wrapper on an LOM. It's broad and horizontal. It's anything that was made available in training data from the public internet. We refine what we do to the publisher's content so it's specific and targeted and accurate. But we're also not trying to drive the I'm your best friend, tell me about your problems, let me recommend a restaurant to you. If you're chatting with a sports publisher, you're not going to ask them for where to go get pizza Friday night, right? But you can imagine over time, maybe that sports publisher does help you find the best sports bars to watch the game, and maybe there's a commercial opportunity underneath that by way of referral or advertising or otherwise. And so it'll continue to evolve. I think if you look at what's happening in the very, very early days around the agent behavior, that also becomes quite compelling and might put a task underneath it, which they're going to be able to do pretty quickly here. But that's not ready for prime time for all types of publishers, the world's biggest, most reputable brands just yet. Yeah, yeah. And so a lot of your publishers, the large publishers, they've got lots of different ways of monetizing what they do. Some of them are subscription based. You mentioned like ESPN and so they've got like TV channels and subscriptions and all kinds of stuff going on.