About ten years ago I went with text-based graph description languages and haven't looked back[1]. One application of AI I have found useful is in the quick co

About ten years ago I went with text-based graph description languages and haven't looked back[1]. One application of AI I have found useful is in the quick composition of graphs via text-based graph description languages: graphviz (the great granddaddy), mermaid.js, plantuml, and of course derivatives like WireViz and CircuiTikZ. A typical ask would be something like "generate a mermaid network diagram showing the relationships of all markup languages". It's surprisingly good. A neat trick you can do if writing in Asciidoc: conditional directives are pre-processor directives, which means they're interpreted before the graphing engine goes to work. So you can make art that has sections which can toggle on and off based on document variables (the Asciidoc lingo for these is "attributes" but using that word can be confusing for old XML hands). Very neat when you have a block diagram you want to reuse for lots of different configurations. [1] Unless I need 3d, in which case I'm afraid I like Blender a lot. Even more than Asciidoc.