I love the recent updates to Shapr3D, and the AI Generative Render tool is a nice touch for visualizing projects. However, I’d love to start a discussion about what I see as the next major leap for this app: integrating a top-tier AI like Claude to help us build the actual 3D geometry.
In my day-to-day workflow, I design and prototype a lot of custom 3D-printed mechanical parts—specifically master mortise jigs for hardware like Emtek pocket door locks and Simonswerk concealed hinges. I’m a visual guy, and while I enjoy the manual CAD process, there is a lot of repetitive setup.
Imagine having an AI assistant in the sidebar where you could simply prompt: “Generate a parametric base plate 150mm x 50mm x 20mm, and cut a centered pocket mortise 120mm long and 15mm deep, optimized for 3D printing with PETG.”
Instead of just generating a pretty picture, Claude could interface directly with the Parasolid kernel to lay out the base geometry or alter dimensions on the fly. You could even feed it a PDF spec sheet of a hardware piece, and have it instantly generate a negative boolean shape for you to use in your design. It would take the grunt work out of initial setups, leaving us to fine-tune the tight tolerances and export the files to our printer fleets.
Currently, Shapr3D’s AI is strictly for rendering and support chatbots. Has anyone heard rumors of them moving toward AI-assisted solid modeling? If Shapr3D integrated Claude to actually draft and modify STEP/STL files, how would you use it in your workflow?
No, For the love of all things holy NO. Just ask any serious software engineer worth their salt what they think of the Vibe-coders. We don’t need vibe-modelers. JUST NO
I am not in favor of this. AI breaks a lot of things and has a lot of security vulnerabilities. Claude doesn’t even get basic facts right a lot of the time. It will agree with you on your broken project and terrible ideas. I’m not even a fan of the AI chatbot they have here.
Pourquoi ne pas vous créer une bibliothèque d’objets avec vos propres fichiers, et les importer ensuite dans votre projet en cours. Personnellement, depuis l’arrivée de la fonction importer projet, je le fait régulièrement et ça ne pose aucun problèmes. J’ai constitué ainsi ma propre collection de standards, pièces simples ou complexes, visserie et ce don j’ai besoin.
If you want to test out AI generated CAD models, there are dedicated software that can do it.
Or you can ask one of the LLMs to generate Open S-Cad scripts for you. I have only tried it with Gemini though.
What you’ll like learn pretty quickly, is that AI is not there yet. You just can’t trust what it does. It’s great when it works, but often it doesn’t. And how do you know that is???
Reminds me of Tesla’s FSD shit, that has just been released here in Denmark. Based on the Dutch work. It works great 99.x % of the time. Are you gonna catch it, when it’s that last 0.x %??? (apologies to all you Tesla/Elon fanboys out there…)
There are 2 sides to this IMO. AI can help save time and resources when used properly. AI can be dangerous and reckless to depend on. I am on the fence about it.
Many coders are against AI because they are terrified they will be replaced. Sorry folks, too late. It’s coming.
I see a lot of arguments about it being “theft” because they train it using people’s works. Well you and I train the same way. Kind of a lame argument.
I was given a project to design the controls for the first (possibly only) titanium injection mold machine. I told them I could use a PLC in the system. I figured I could learn PLC programming. That was a HARD “NO! NO COMPUTER CONTROL”. Why? Because the last project they built was a steel stud manufacturing machine (similar to a roof gutter making machine). The programmer they hired held them hostage towards the end of the project because he wanted more money.
I am “vibe coding” now. I don’t have the time to do all the work so I am using a tool to help me. *Claude. Could I write the code? Sure… in days or weeks. Could I hire a “coder” yeah probably no. Maybe later if I monetize a project and get something going. But right now it’s helping me solve problems.
AI is here to stay. It’s actually been around for a lot longer than most people know.
I will say that I want AI to be an Appliance that I control and not controlled by corporations. One that works for me. I think that might be coming. I hope.
Change is always hard.
Will I use AI to design in Shapr? I cannot see it working that well. Coding is one thing, advanced 3D design? Yeah not likely on home PCs and $20 a month AI.
I’m sure industry is leveraging it heavily and likely won’t admit it.
How many engineers and at what expense do you think it takes to make 1 Samsung cell phone? 1? 10? 40?
Sooner or later Shapr may have no choice but to adopt it to stay alive.
Completely agree and support, Fusion already allows this and it would help a lot. If someone does not need, good for them. If it helps someone on their flow, great. I was debating which tool to use and this is just the final straw for me to go to Fusion.
I don’t intend to ever use AI, I actually enjoy making things myself.
I’m not against including a Claude pipeline into Shapr3D, I’d just hate for it to be a higher priority over other frequently requested improvements of the software. Do it for all I care, but more fundamental functions and additions should come first IMHO.
One thing I’ve learned in trying to use AI in design is it doesn’t generate files compatible with CAD apps or even STLs. I also don’t trust the AI I’ve worked with, mainly ChatGPT. At least half the time, I have to correct it.
Why? Because the last project they built was a steel stud manufacturing machine (similar to a roof gutter making machine). The programmer they hired held them hostage towards the end of the project because he wanted more money.