Yes. And Stephen, I'd like to just give a little bit more perspective on some of the investments, particularly where we're heading on AI, generative AI. And I think it's important to see that. We see it as a strategic enabler of what we are trying to accomplish, which will really drive more affordable, accessible and personalized care. We've been embedding AI responsibly, not just as an experiment, but at scale. And we see that they're improving our efficiency and our outcomes and also the experience for members, providers and associates. And maybe just to put some real tangible examples on that, I shared a little bit in my opening comments. For members, for example, our personalized match feature in our Sydney helps 1 in 5 of our members right now, select the right provider using more than 500 personalized data points and improving navigation and satisfaction. We're using it across our customer service. We have tools that improve our first contact resolution, shorten wrap-up time, help with proactive engagement. And I think really importantly, for the commitments we've made on care providers, our HealthOS platform is automating our onboarding, our contracting, our roster management. It's also reducing a lack of information so that we have reduced denials by more than 68% and peer-to-peer reviews by over 100%. So we're getting real-time data. That's also giving us greater insight into some of the things that Felicia shared, which is how we get ahead of the cost curve. So across the enterprise, we see it as a huge opportunity to help support our productivity goals. We look at it to reduce the burdens on care providers by reducing our chart request by almost half. And then our national account teams, for example, are using it to update benefits and onboard clients. So across the board, including for our own associates, we're investing in them as well. We just signed a partnership with OpenAI, where we're going to actually train our folks to be able to use these skills appropriately. The reason I wanted to share more detail on that is we're embedding these at scale across our operations and prioritizing high-use impact cases that reduce, first, the complexity, they drive savings and enhance the experience, which are 2 of our core goals: one, reducing the cost curve; and two, enhancing the experience our members have. So what we see is this is going to create leverage for us, improve affordability, strengthen operational performance and support sustainable long-term growth. And to your last question around sort of the impact of those, we see these as front-loaded investments across the board. Mark shared that. We're seeing great pickup in the investments we've made in Stars. I just shared the AI investments, and we think that they help support us. So thank you for that question.