I delve into the world of Product Lifecycle Management, CAD, CAM, Product Data Management, and the like a few times a year. A few supplier companies have invited me to their user group conferences over the past few years. A recent interview was with the new CEO of a cloud-based PLM product who also talked extensively about the coming expanded use of Artificial Intelligence or AI.
So many press releases come my way touting AI that I wonder if it’s only a marketing term. How much is real…and how much is hype?
One thought leads to another, and I wound up on a Microsoft Teams meeting with two people from PTC. Not unlike other companies, PTC discovered it would be much better to acquire a cloud-native company than to try to reinvent the wheel. Enter Onshape.
Last week Darren Henry, SVP Onshape General Operations and Cody Armstrong, Sr. Director Onshape AI Innovation discussed the benefits of cloud-based software (something I’ve written about for years) and how AI is far more than marketing hype. I offered Cody that challenge, and he stood up to that challenge. This is a longer than average post for me, but it was a longer than average conversation 🙂
I’ve had conversations with people from PTC, but there has never been one with Onshape. So, Darren provided an overview. Like I mentioned, I don’t need to be sold on the benefits of cloud-based software for manufacturing. It allows enhanced version management, change tracking, and scaling.
He offered the example of a company sending drawings to selected suppliers in order to obtain quotes. When you send the file, well, they have the file. Can you trust what they will do with your proprietary intellectual property? If you allow sessions on the cloud platform with restricted permissions, then you have more control over the process. Of course, the winning bidder will use the latest approved drawings for building the parts or fabricating. And everything is traced.
Switching over to Cody, whose task was to convince me that AI is real, a useful tool for designers.
“But the first thing that would point out is AI-powered design assistance. We call it AI Advisor. We believe it’s an industry leader. It’s available to millions of users today and the reception has really been fantastic.”
It’s a Large Language Model trained not on the world at large but on the company’s own database and standards. So, like LLMs you may be using now, he adds, “You can ask questions inside of Onshape about how to use Onshape and it will give you a tailored answer to that response. And that answer is much more accurate than what you would get with ChatGPT or any other traditional LLM.
“We’ve built our own layer on top of it that augments that data with our own learning materials, essentially defines it as a source of truth. And so we believe this is the new way that people will learn to use Onshape. And over time, you know this will become more and more.”
Another benefit of AI Advisor is training. “One of the things that that we’ve often found with parametric CAD in general is that it’s very difficult to learn, very difficult to use for a new user. Even if you have experience with a CAD system, learning a new CAD system is difficult. And so what we really set out to solve with the AI Advisor is the ability to quickly answer a user’s question, especially as they’re learning.”
From here, we got into previews of coming attractions.
Another cool use of LLMs comes from language ability. “Another big benefit of A I is we get localization and translation as a tool of the LLM. And many of these localized LLMS support many different languages. We can support those languages as a byproduct of that. So we’re really excited about being able to answer questions in any language, including languages that our product doesn’t even support.”
The next cool tool is FeatureScript. Cody continued, “The next topic that I wanted to bring up is FeatureScript autocomplete. FeatureScript is a language that that we’ve developed for building CAD geometry. It’s very like JavaScript like. You build custom features. They allow you to take all these tedious tasks and turn it into a single feature. It just allows you a quick, easy way to create a feature that’s tailored to you.”
A side benefit—it enables an average design engineer to do software development, which accelerates development.
Search has been around for decades, but it doesn’t seem to be getting better overall. Cody discussed Onshape’s coming use of AI-powered search. “We will initially launch this hopefully in the Public Space. They have a library of of millions of models that are publicly available that users have created using Onshape. We want to allow users to quickly find information in that giant list of documents. We think A I is really well suited for that.”
The AI-powered rendering engine is cool. As Cody explained, “This is a topic I think a lot of people will really appreciate. Rendering can somewhat be a tedious, time consuming task. It can often take 30 minutes to an hour to generate a good rendering of an object in CAD.”
The process begins with the CAD model. You incorporate the background and realistic materials. We can generate professional grade renderings with nothing but a simple prompt, right? And using the latest generation A I image generation models, we can really get amazing results in seconds as opposed to 30 minutes or an hour in more complicated. We call it AI quick render because you can go from a model to a rendered image in a background that looks good in seconds, right?
The last thing I’ll mention, and the furthest down the road for development, concerns AI Agents. They consider agents as simply part of the team. I recently attended a conference where the speaker mentioned a future release containing Model Context Protocol (MCP). The “influencer” sitting beside me almost gave me a bruise on my thigh hitting me with abundant excitement.
Well, Onshape is working on their MCP application.
“So it becomes a tool for productivity that the company can benefit from. But importantly, our architecture will allow for this,” Cody explained.
“It will allow us to say this is a company specific agent with specific permissions that only certain users can access. We we scope that in a very narrow way. It’s a toolkit for other companies to build Agentic workflows into your software. So if you’re a company using Onshape, and maybe you want to build your own agents, we want to allow and enable that. This will allow companies to build their own agents and set their own definitions and communicate with Onshape through MCP interaction.”
Click on the Follow button at the bottom of the page to subscribe to a weekly email update of posts. Click on the mail icon to subscribe to additional email thoughts.
