The AI is a cool idea. Maybe as an add-on? An import? A library? Collaboration? IDK, but details like via tenting properties on layer 3 are best left to the experts.
We see this in the web browser engine space with Chromium.
That's why I find the glee with which prominent EE community types appear to condemn Flux really frustrating. Yes, Flux has big problems and a huge mountain to climb. To me, that's why we should encourage them. There are a million easier things those folks could be attempting, so shitting on them for trying to bring a new perspective is self-defeating.
As for Altium, I haven't used it because it's very expensive and tied up in a larger enterprise ecosystem. It doesn't make a ton of sense for solo or small-team designers to use a tool heavily influenced and full of tooling for use by the world's largest companies.
That's why I found Flux so compelling; my partner and I actually want to use the "Figma for electronics", if such a thing can be provided without critical bugs.
The tools need to work consistently, just like the tools of a carpenter. If his hammer head keeps falling off, it's not very helpful even if it was free.
My point was that it is really difficult to make sch and pcb design easy for the end user. There are already companies that have climbed that mountain. AI is a nice differentiator, but in my opinion a startup would be better focused by working with existing tools instead of reinventing them (poorly).
Obviously startups are free to do as they wish.
The argument I see for Flux is the "FAE in a box" sorts of features for educating hobbyists and early-career design engineers while keeping them out of trouble. The person I imagine would otherwise not done a PCB at all. Someone who instead would have wired up up a bunch of prefab Arduino-type dev boards or bought small power supplies, DIN mount relays, and so on. Those features put a lot of things within the reach of technicians, panel designers, and systems integrators.