Recreating Apples scroll based interactions is fun 👀

I've always wondered, how do you recreate Apple's iconic scroll-based interactions? Well, I just did it using Framer's agents in less than 30 minutes and without wasting a bunch of tokens or reprompting agents over and over again, thanks to their efficiency updates, which made agents complete tasks in 15 to 30% fewer steps and using 30 to 40% fewer tokens. And you can try it yourself by copying my prompts using this link. But basically, I prompted my agent to create a scroll sequence component that plays through a video or image sequence frame by frame when someone scrolls it. Then I went to Higgsfield to generate a short video using Minimax H3 with our Flowglide coin reference image images attached. Then I uploaded that video text field to Framer, and voilà, that's how I created this scroll-based interaction, which is nice because