Ran across this and was wondering, who is this geared toward?
AI creation of standard typical designs of common parts. Their tag line is real stretch “Get the full power of a professional CAD platform”. Haven’t seen anything on drawings or how you assemble parts, manage the data.
I downloaded and tried it. Everything I asked it to do failed. I asked for a bowling ball and got a cylinder. I asked for a boat and got a hollow box with a lid on the side. This kind of thing will eventually work. It will probably be aimed at people who do a lot of repetitive things, like bolted beam connections, or machine design elements. That’s my guess anyway.
I thought about the repetitive design aspect, we do a lot of “per order” variants but its based on standard design templates that need to be revision controlled and industrilized with our factory. Just seems easier to copy the latest and then make the edits.
The bolted beam connection for example, the AI has to be smart enough or trained enough to know not change certain standard parts. If I have to check everything it does, then its faster to copy/edit.
What would be more helpful is controlled edits with AI:
“Change all plates to 3/8 thick galvanized steel.”
“Change mounting bolts and attached hardware to 1/2”
Yeah and that means the A.I. needs to somewhat be customized to the ‘lingo’ inside the company so that the A.I. does not refer a flat bar as a plate for instance.
Like the 2000s IT bubble burst, the AI bubble is going to end at some point.
This is the venture capital economy “side issue” (a tragic one imho) with literally billions of USD poured into lot of silly or super hyped IT projects that are not going to deliver or will be utterly incomplete or far from usable.
Easy money with surreal returns in the short period, less investments into “real” manufacturing.
Company know how is not completely digitalized and quite incomplete inside the IT field and sometimes marketing BS makes its way into the technical information too, poisoning the data scraped by the AI agents. A huge pile of BS made even bigger.
I had an IT professional sending a script full of hallucinations to solve a problem that cannot be solved with a script in the first place: using Copilot and it was about a stupid windows server issue. (a microsoft AI agent should be knowledgeable on MS stuff, but let’s ignore that little detail for a moment)
Once the AI takes over the medical field, you will get every kind of diagnostic filtered through AI. All good unless the training set is wrong or the medical professional starts relying 100% on AI without any critical thinking.
CAD need answers that’s 100% correct.
AI is still “fuzzy”. I don’t think AI got any 100% correct answers.
It’ll be “funny” when AI making medical cocktail to help patient.
“Oh you have a cold and sore feet? Mix this two and drink it twice a day.”
AI need a setting for 100% “true” info and can’t give results that contradict them.
On the other hand AI is great at making funny photos.
More funny is in their website.
User need to pay for “Text to CAD” credit to talk to the AI. Another subscription.
Remember old style paid chat on telephone?
The Internet has also created a lot of improbable markets that in the past would be limited to some small niche. Telephone chats were indeed a thing. I never grasped the concept to pay for something like that tho…
I see some very suspect stuff in this video. He searches for a wheel with grooves in their company data, the result is returned, but the model is already in the tree, just hidden.
Probably a promotional video to get interest and investment.
Getting AI to search internet is the easy part.
Getting AI to model in SW or any CAD will be difficult.
Whoever get it done will be rich.
Last thing, its not a violation of ISO 13715 to have sharp edges in the model. the standard just tells you how to create callouts on the drawing for edges, whether a burr or undercut is allowed. This is covered as a note/symbol on the drawing, we wouldn’t model it.
And didn’t the fork have chamfers? I mean, they look like chamfers to me unless his model shading lighting is off. And he said remove the fillets, not chamfers?
Could be, unless what @Frederick_Law said about trying to get investment, they just throw together something canned to show what they expect it will do when finished. Still a bit shady IMO.
According to Linked In, they have investers and have received 3 million so far.