Yes. So just to give you a quick backdrop, I mentioned at the beginning of the call that generative AI is the next platform shift in technology. A matter it's in technology, it's actually in society. And if you think about the -- when these ships occur, like the PC or the arrival of the PC and the PC to the web, and the web to mobile, right, you can see the kind of disruption that occurs in market after market when these transformations happen. In fact, it's interesting, there was a Wall Street Journal headline today that says, is this the iPhone moment. And I think, absolutely, I would answer yes. Yes, it is. Now there's an interesting question though, if you remember in the early days of, say, the web, where companies were trying to figure out what do we do? Do we give a brochure on the site? Remember when companies used to say, like we're not allowed to like off of our site because there's a legal problem of that? Everyone’s like, what? So people have to figure out how to use these new technologies in the corporate setting. And I think that's what the conversations I’m having with customers now are exactly that, right? Is this ready for an enterprise use case or is a bot that I put in front of my customer is going to start like talking saying stuff that I don't want to say, right? Is it going to start having a dialogue with my customers about God knows what -- or are they going to stay on topic and can talk about my products, my services and all that kind of stuff. And so I think that's where a lot of the work is going right now. And I think there's really good questions that are getting answered every day that work we are doing, work others are doing in terms of like how to keep these large language models on topic and provide boundaries for them so that they are useful in a corporate context. And that stuff is getting resolved, I think, pretty quickly. And so the converse -- I'll put you into a conversation now with the customer recently which I think is indicative of what I think is going to happen. It was I was talking to a customer, a very large financial services company. And they were telling me how they had spent the last seven years building out all of the intent to have a bot for their service use cases that could contain customer culture containment as the -- didn't have to reach a person. And they said, well, this containment was after seven years of work or whatever, it's about 40% and so 60% of the call is made through that. And I asked -- and we're talking about large language models, and I said, do you think you're going to keep that investment or do you think you're going to start from scratch in the large language model world. And the customers say now, I will keep that investment, but hopefully, large language models will help us move it forward from 40% up from there. And through the course of the conversation, we talked a lot about what's possible in the architecture of these new language models and how they can work with segment customer data and things like this. And at the end of the conversation they asked again, do you think you're going to keep that 40% -- the investment you made over the last seven years that got you to 40% containment? And the customer said, no its going in the garbage can, right? Like every decision we've made for the last seven years about what's possible is now like a relic of the past and we are up for relitigation and potentially new approaches, new vendors, new ways of implementing it because large language model world just upends what is possible. And I think that is why it is a gift in terms of creating new opportunities for companies like Twilio who is helping our customers to activate their customer data across the customer life cycle, take CRM, which has historically business like kind of a sleepy area of like just a database, not activated, making useful across many different touch points. Large language models are an absolute gift, and I'm very happy that we bought segment when we did because the data that is in segment enables a company to customize these interactions based on who they're talking to. The end user, the customer of our customer, and that is very powerful. So anyway, this is day zero of large language models, and you'll be hearing more from us in the course of this quarter. Obviously, we have signaled in August and I would not be a responsible technology leader. If AI wasn't prominently a part of what we're talking about at Signal, so we'll have more coming, and I hope everybody joins us in Signal in August.