I've been watching Jake do the Headless Frontier work with two different AI bots -- ChatGPT and Claude.ai. And as he's doing that, I'm slogging away the same way I've always done it, working on the top level user interface of WordLand in browser-based JavaScript. I don't see a way around it, because I have a special way of working on user interfaces, and we're still quite a ways away from the bot being able to do vibe coding at that level. It's fascinating to watch Jake revive code I wrote in the late 80s and early 90s. He's a very accomplished user of it, being transformed, with the help of the bots, into a kernel-level developer of what's basically an OS built around a scripting language, object database and with the internet latched on after the whole thing was done, and then ported to Windows. I stopped working at that level before all that michegas happened. I have looked at the code Jake is working on to see what became of it, and wasn't horrified, I recognized my work, but I wouldn't ever want to work on that myself. I imagine some commercial developers have already rebuilt their testing and support functions for products around ChatGPT-like systems. If you're an old Frontier fanatic, that's where our product is once again getting out in front. When Jake is done it'll be one of the first big systems totally managed in an AI system. It should be relatively easy to add new verbs to the language, even to add new features to the language, new APIs, etc. #