 The company announced Lumiere. It's an AI video generator that it's calling a space-time diffusion model for realistic video generation. As far as text-to-image AI video generators go, pretty impressive, at least from what they've shown us. Does Lumiere feel like a leap forward? They have announced something, they've shown us videos, they've promised demos, but we cannot exactly get our hands on it. Google's Lumiere is supposedly doing, which again is contextualizing here is the scene, here is the subject in the scene, here's how that subject needs to get from A to B. So instead of it interpolating and trying to make the frames bleed and mesh and melt into each other, it's thinking about how those two interact. And that is revolutionary. Gavin, my co-host on AI for Humans, we pointed a little bit to the bleachers and said, we think in two to three years time, text-to-Hollywood level content is going to be possible. And I think we're on that trajectory still.