Generate not only GC scripts from text input, but also geometry through a combination of text and graphical input. For example, starting from pre-drawn 2D or 3D geometries, explain in text how they can be automated, modified, and distributed. Complete visual programming through a text-geometry-text-geometry process that leads to the final drawing. I think Generative Components is the software that is closest to the AI ​​approach. Therefore, automate Generativecomponents with AI and then merge it with Microstation (or verticals), or both.