There’s a lot of focus on making AI models bigger, faster, and more capable. I mean, yeah that clearly improves what they can do. But the more I’ve been working with them, the less it feels like capability is the bottleneck. It’s really about the context layer. Right now, you don’t really see how the model is interpreting what you give it, what it keeps, what it drops, or how it connects things. That stuff is mostly hidden. You can nudge it, but you’re still operating inside something you have...