AI CAD sounds awesome… until it's not.

There’s this big idea floating around right now:

Sounds incredible, right?

But here’s the truth:

Most of us are awful at describing our designs.

I got a message recently that said:

Uhh… what?
What’s a follower? What’s beneath it?
Where are the dimensions? A sketch? Anything?

I mean, I’ve been doing this forever, and I still have zero clue what that model’s supposed to look like.

And look — I’ve played around with a bunch of these new AI model builders.
​

The one that feels the most promising is Zoo.dev (used to be KittyCAD).
Great concept.

You describe a part, it builds a parametric CAD model.
That’s exciting stuff.

Buuut the few times I’ve used it?

Nothing.​
Literally no model...

Maybe you’ll get lucky — maybe it’s working this week.
But from what I’ve seen, it’s still very hit-or-miss.

Now, when it does spit something out, you’ll get a dumb solid.
No timeline. No parameters.

But that’s not useless.

Drop it into Fusion, and you can start doing your thing:
pull faces, stretch holes, delete weird stuff.
That kind of hybrid workflow?
Honestly, I’m into it.

What I’m really holding out for, though?

Reverse engineering from photos.

If I could snap a few pics of a part, toss in a couple key dimensions, and get a clean parametric CAD model back?

Total game changer.
I’d use that as the base, then design all the custom stuff around it.
Enclosures. Fixtures. Mods. You name it.

Or I design the base model and then ask the AI to add fillets everywhere, a thread on a cylinder and centered hole.

I would love this HYBRID Approach!!!

We’re not there yet.
But man, I really want to get there.

Until then?

You’ve gotta know how to build solid CAD models.
Clean sketches. Good parametrics. Thoughtful design intent.

Because AI isn’t replacing that anytime soon.

So — I’m keeping the Jumpstart Course and others on sale for the next 24 hours.

​

​