Jim Brown had the opportunity to attend this year’s Autodesk University in San Diego, CA. The conference was well attended, with approximately 12,000 people in total. The mood was optimistic and there was lots of vibrant conversation all around.
General Impression
This year, the conference that we all know as “AU” felt a bit like “AI-U” as Autodesk highlighted their progress with artificial intelligence (AI). Pretty much every briefing or conference we go to these days highlights #AI in some way. But #AU2024 went beyond the others, perhaps with good reason, because Autodesk is really putting their money and some top-notch talent where their mouth is. There is some real substance to what they are doing.
But AI wasn’t the only focus. We were also able to hear about Autodesk’s role in helping prepare the venues for the upcoming LA28 Olympic Games. We left with takeaways in four key areas:
- AI Research and Plans for the Future
- Currently Deliverable AI Enhancements and Applications
- Progress on Autodesk Platform Services (and Fusion)
- Continued Investment in Core Products
Yes, AI counts for two of our four key takeaways. We feel that represents the focus of the conference. Let’s spend some time on each.
Autodesk Goes Big on AI
All software vendors are focusing on AI to some level. Although AI is not new, generative AI and large language models (LLM) have made applying AI more visible and accessible, primarily due to the popularity of ChatGPT. The use of AI in enterprise and engineering software is now an expectation and vendors are spending their efforts making sure their customers – and the market – know that they are working on it.
Increased AI adoption has been an evolution and will continue to progress over time, but some believe that we have hit a tipping point that will dramatically change the way people work with technology. Autodesk clearly believes that AI will fundamentally change the value delivered by their systems, and used AU as the platform to share their vision for how this will unfold. As CEO Andrew Anagnost shared in his keynote, although the industry has been in a period of hype, many companies have moved beyond experimenting with AI to investing in it. He also shared that he’s seen a level of frustration on customer’s part as they try to find the value. But, he explained to customers, “the great sorting out of the good and bad is just beginning, but I don’t think you should have to figure that out yourself,” he explained. Autodesk sees their role as researching and developing the technology and use cases that will drive customer value on behalf of customer to achieve mutual value.
Autodesk’s AI Strategy
#AutodeskAI, “your AI-powered design and make partner,” is the center of the strategy. Autodesk is quick to point that AI will expand what people can do, not replace them. Their strategy is for AI to “Automate, Analyze, and Augment,” removing non-value-added tasks and allowing humans to achieve more by focusing on what they do best. As Autodesk CTO Raji Arasu shared, “Your value is not just creating geometry – it’s creating ideas.”
To help AI empower users Autodesk is working in two directions. The first is through top-down research projects like “Project Bernini” that promise to change the paradigm of how humans interact with systems, and in particular with 3D spatial data. At the same time, they’re working from the bottom-up to improve existing work methods and paradigms. We’ll talk about current, deliverable AI capabilities shortly. At some point, they say, these efforts will meet and work together. They are investing in what can be done today, but also trying to build knowledge and competency for the future.
Say Hello to Project Bernini
Let’s start with Project Bernini. Autodesk’s R&D investment is impressive. Autodesk was early in the use of AI to generate design concepts, an approach known as “generative design.” Since that time, they have continued to push the limits on research and Andrew Anagnost has not been shy talking about it. While most research today is focusing on large language models, Autodesk is developing approaches for physical and spatial AI. Text-oriented AI is valuable and becoming more commonplace, but Autodesk recognizes that they have a unique opportunity to focus on 3D to further their “design and make” mission. This is the purpose of their research project, Bernini.
Project Bernini is #GenerativeAI for #3D shape creation. According to Autodesk, Bernini was trained using the largest dataset of 3D training data ever assembled. They use that data to train a spatial, geometric AI model that you can prompt in a variety of ways. Beyond text, like most AI inputs, you can prompt Bernini with 3D data like voxels or point clouds. From this, for example a rough sketch, Bernini can generate geometry. The idea is to augment the designer so they can ideate quickly, get to the first stage of 3D modeling, and move into the next phase of design.
Project Bernini is not a product and is not destined to become one. It is a proof of concept. This is because the license for the large 3D dataset they used for training doesn’t allow commercial use. I asked Andrew about that in the executive Q&A and he explained that while the model trained from the dataset isn’t something they can use, what they learn from it is. Autodesk is employing a team of PhDs to do the hard data science to develop methods to train foundation models based on 3D data. That intellectual property is what they will build into future product offerings. It’s a long-term investment strategy and they are betting big.
Current, Deliverable Value from AI
Project Bernini is the big, top-down bet, but only a part of what Autodesk is doing. It’s still early days for most manufacturers to expand beyond core AI capabilities. Instead, they are looking for practical ways to get their work done. This is what Andrew refers to, in a positive way, as “boring AI.” It is artificial intelligence that is ready to deliver tangible value. Autodesk Executive VP Jeff Kinder explained that AI doing “boring, practical things” will help build trust and allow people to see the value of AI.
It’s comforting to see long-time Autodesk design and manufacturing veteran Stephen Hooper taking an active role in defining their AI strategy. It’s also another proof point of the investment Autodesk is making in AI. As Stephen reminded attendees, they have been preparing for AI for the past 10 years as they’ve moved data to cloud and made it granular. Now, he says the investment has been made, the hard work has been done, and we should expect to see Autodesk accelerate their AI capabilities for design and manufacturing. They’ve already made strides in generative design and will focus on automation to allow people to focus on creativity by allowing them to move beyond repetitive, time-consuming tasks so they can focus on innovation.
Progress to date includes the Autodesk AI capability to generate sketch constraints that will cascade through the life of the part, eliminating a common source of 3D modeling errors. Instead, AI in Fusion will automate the creation of constraints and ensure they are correct. In addition, Fusion Drawing Automation will automate and streamline manual tasks related to creating 2D drawings, like removing fasteners and automating dimensioning in the appropriate format. This is time-consuming, repetitive, and non-value-added work that can now be turned over to AI agents.
Other progress has come from acquisitions, including the new Alias Form Explorer, based on their acquisition of BlankAI , which creates native Alias geometry to spark innovation. In addition, Autodesk announced they are acquiring NAVASTO for physics-based simulation and CFD. We are told it will be available in Alias next year.
Progress on the horizon includes the Autodesk Assistant, which is coming soon to the Manufacturing Cloud. It’s a context-aware agent architecture designed to perform tasks. Autodesk will create Autodesk Assistant capabilities, like the ability to respond to a request like “rough a part” by providing a toolpath strategy, or answering questions about manufacturing using information that’s in the Autodesk knowledge base. We expect to learn more about this in the near future.
It’s important to mention Autodesk’s focus on AI governance, privacy, and transparency. This will be crucial to Autodesk’s ability to effectively take advantage of its AI investment. For example, Autodesk was clear that they are using generalized process data for model training, but are not using any of their customers’ design data to protect their intellectual property. To help companies understand how AI uses their information, Autodesk explained that they are working on something similar to nutrition labels to disclose how they train their models.
Kia Motors, a Case in Point
We heard from KIA Motors Senior AI Designer, Voho Seo, as an example of the potential of AI in design and manufacturing. He explained how they use design data from past projects for future designs, accessing their existing datasets to train AI. He explained that they are working with Autodesk as a research partner because they wanted a partner that knows both AI and design. He shared three learnings from their AI experience:
1 – Start using AI like onboarding a team member, you need to know your processes
2 – Understand AI’s strength, which is using text and images to guide AI to come up with new ideas
3 – Recognize that human insights are the most crucial aspect and that you can’t let AI design everything
Platform Progress
I realize I’m only two bullets into a four-topic article, and all I’ve really spoken about is AI. I think that sums up my conference experience. But AI is not what customers are using today, and it was important to understand the progress that Autodesk is making with Autodesk Platform Services, formerly Forge, and the manufacturing representation, Fusion, which is also known as the “manufacturing cloud.”
Stephen Hooper shared Fusion’s progress around data, including the ability to connect data from the desktop to the cloud or between cloud applications in a hub-and-spoke model. To support this, they’ve exposed the data model with more granular APIs. Stephen gave several examples of how companies have used the new manufacturing data model API to connect cloud-to-cloud with other solutions including:
- Integrating with Paperless Parts to quote directly from a Fusion design
- An Avnet add-in for cost and procurement directly in Fusion
- A Makersite plugin for Inventor to calculate sustainability metrics
Shelly Mujtaba, VP of Product Data, shared more details about the core capabilities of the manufacturing data model, specifically “Granularity, Interoperability, and Accessibility.” He explained how breaking data into smaller, digital parts will help manufacturers overcome the inherent limits of file-based collaboration. For example, he shared how granularity and openness will allow manufacturers to enrich their data, and access their data without the need to open the authoring application to access core information. He also explained the value of building on the Autodesk Data Model in order to enable cross-industry use cases with applications from architecture, engineering, and construction and media and entertainment.
Core Products
Beyond the platform, Autodesk VP of Industry Strategy Srinath Jonnalagadda shared the importance of continuing to invest in their core products, like #Inventor and #AutoCAD, because they are the “mainstay” of their current business. I was impressed by the list of 140 functional enhancements, including a focus on reliability and industrial scale assemblies, in a theater presentation on Inventor improvements. In particular, as a former product manager, I liked the fact they dedicate capacity to fix issues reported by the user community, what they call their “JDI” or “just do it” enhancements. These may not be strategic on an individual basis, but in aggregate help with user adoption, satisfaction, and success.
Autodesk University also included a day-long PLM Summit. Autodesk VP of Cloud Data and PLM Derrek Cooper shared that #Vault and Fusion will continue to get closer together, and that Autodesk wants to deliver as much as possible out of the box in an integrated fashion, but will also remain open to connect with other systems. This includes using Autodesk Platform Services and the common data model. There were some interesting announcements, including the availability of a synchronization function between Vault and Fusion Design. This integration is intended to enable downstream functions in Fusion to leverage product data from Vault. They also discussed the Vault API designed to take data from Vault and bring it into other cloud applications (like Microsoft Teams).
There was much more information about core products at Autodesk University and the PLM Summit, but it was simply too hard to get to everything in the time available. Autodesk University was a packed and highly valuable experience, as usual.
Wrapping Up a Long Post
My note is a little late, but I hope you find it helpful. Thank you to Jason B. Love and all of the others for the invitation and coordination that helped me get the most out of my AU2024 experience, and thanks to Autodesk University staff and Autodesk leadership for making the time valuable.
Tech-Clarity is an independent research firm dedicated to helping industrial companies learn how to improve performance through digital transformation. For more vendor insights and research on the business value of technology follow us or subscribe.