We're looking forward again this year to sponsoring the Global Product Data Interoperability Summit (GPDIS), where the overall theme is AI Enabled Interoperability Across the Digital Enterprise. The event takes place from September 15–17 in Phoenix, Arizona.
At this time when even our dishwashers and doorbells are online, it is hard to recognize that there are still some aspects of digital technology that has remained the same since the dawn of the computing era.
Back when mainframes took up entire rooms, screens were black and white, and data was entered using punch cards, it didn't take long for a fundamental principle of computing to emerge that remains equally true today: garbage in, garbage out.
This principle takes on a new dimension in the still-developing structure of AI-informed product design and manufacturing. It doesn't help anyone if data being aggregated by an LLM is inconsistent or inaccurate. Maintaining 3D data fidelity is part of what I will be addressing in the presentation I am giving at this year's Summit: Assuring AI-Ready High-Quality Product Data Out to the Far Edges of your Enterprise.
For all the insights potentially to be found by having machine learning systems crawl through the product development lifecycle, they are only going to be as valuable as they are able to access accurate data. Within the corporate office park or server network of a major OEM, the challenges of maintaining the digital thread can be reasonably contained. It is the inherently multi-stage process of manufacturing where ensuring the accuracy and continuity of all data associated with the model becomes far more complex. This is area for greatest risk of "garbage" getting "in."
During our presentation, we will be talking about how the latest standards for STEP and other formats like QIF can help keep product data clean and traceable. As I state in my abstract:
This session explores the often-overlooked challenge of extending high-quality product definition data beyond core engineering systems into the extended supply chain—where variability in tools, formats, and processes is the norm. While Model-Based Definition (MBD) promises a fully digital, machine-readable product definition that maintains the digital thread and reduces human interpretation errors, achieving this at scale requires robust validation, governance, and interoperability across diverse environments.
Drawing on real-world experience enabling suppliers and OEMs, this talk will present a practical framework for ensuring AI-ready product data—data that is not only syntactically correct, but semantically consistent, traceable, and verifiably aligned across native and derivative representations (e.g., CAD, STEP AP242, QIF). We will show how data quality assurance must evolve from point checks to continuous validation pipelines that span design, manufacturing, and inspection workflows.
The information we'll be presenting is based on our decades of experience serving the manufacturing sector; in particular the aerospace supplier market. The aerospace and defense industries have been at the forefront of adopting model-based definition (MBD) practices and for working to help it take hold in their diversified supplier base. We've been working with a focus on aerospace for quite a few years, with both our Validate and Revision products, and look forward to the chance to share these insights as they relate to empowering AI with accurate data from across the supply chain.
We look forward to again connecting with other companies and organizations involved in laying the groundwork for greater interoperability at the summit in Phoenix.
And next time you're frustrated trying to use a smart device at home or at work, just be glad you don't have to keep track of all the punch cards!