The GUI library aspect is progressing well, but getting a desktop environment going is challenging even with AI. Still I want to capture things like the document workspaces and such. Lately Ive been tempted with making a postscript like display server based on GPU accelerated vector primitives.