For years, CAD and AI have lived in different worlds. Engineers designed; data scientists modeled. And while machine learning revolutionized sectors from retail to biotech, CAD data—the lifeblood of design and manufacturing—remained stubbornly inaccessible.
Now, Tech Soft 3D wants to change that. The company, best known as the world leader in engineering software development toolkits (SDKs), has launched HOOPS AI, the first framework built specifically to bring machine learning and AI to CAD data.
The new platform gives data scientists and ML engineers a direct pipeline from raw CAD files to production-ready models, eliminating the patchwork of scripts, file conversions, and brittle workflows that have long frustrated the engineering world.
“HOOPS AI represents a major leap forward for organizations looking to finally harness artificial intelligence for 3D CAD,” said Gavin Bridgeman, CTO of Tech Soft 3D. “It provides a complete, reproducible pipeline that makes machine learning workflows with CAD data both practical and scalable.”
Why CAD Has Been a Blind Spot for AI
Despite powering modern manufacturing, product design, and simulation, CAD data has been virtually invisible to AI tools. Standard machine learning frameworks excel at handling images, text, and numerical datasets—but not the massive, hierarchical, geometric, and metadata-rich structures found in CAD.
Until now, data scientists working with 3D engineering data had to rely on improvised scripts, expensive CAD software, and fragile data exports, often losing fidelity or interoperability along the way.
HOOPS AI changes that by combining CAD access, dataset preparation, and ML encoding into a unified workflow. Instead of juggling formats or worrying about data loss, users can focus on building, testing, and iterating models at scale.
What HOOPS AI Does
HOOPS AI builds on HOOPS Exchange, Tech Soft 3D’s industry-standard CAD import/export engine, which supports 30+ file formats including CATIA, NX, SolidWorks, Creo, Autodesk Inventor, and more. Through a Python API, data scientists can directly access:
- Geometry and topology
- Assemblies and product manufacturing information (PMI)
- Metadata and materials
This approach avoids risky intermediate conversions and removes the dependency on commercial CAD software licenses—something that has traditionally hindered AI research on engineering datasets.
Once CAD data is imported, HOOPS AI automates the next steps:
- Dataset preparation at scale — including cleaning, segmentation, and visualization.
- Encoding CAD models into ML-ready formats, compatible with frameworks like PyTorch and TensorFlow.
- Experiment management, including versioning, dataset tracking, and reproducibility.
Built-in visualization tools provide real-time feedback during every stage, letting engineers and data scientists iterate faster with full transparency.
Why This Matters
The potential applications are massive. Machine learning models trained on CAD data could transform everything from generative design to predictive maintenance. Imagine AI that can identify weak points in assemblies before manufacturing, or suggest geometry optimizations to reduce material waste.
In today’s industrial AI landscape, most innovation happens around text and image data. HOOPS AI opens the door to 3D intelligence, where models can learn directly from the world’s engineering blueprints.
Tech Soft 3D’s deep experience in CAD interoperability also gives it a critical advantage. Its HOOPS Exchange toolkit already powers software from Ansys, Siemens, Hexagon, NVIDIA Omniverse, Unreal Engine, Unity 3D, and hundreds of other engineering platforms. That experience translates into one thing every ML team wants: trustworthy data ingestion at industrial scale.
A Catalyst for AI in Engineering
The launch of HOOPS AI also highlights a broader trend: the convergence of engineering, data science, and AI infrastructure. As manufacturers adopt digital twins, generative design tools, and edge analytics, the ability to train and deploy ML models directly on 3D data will become a differentiator.
For Tech Soft 3D’s 750+ independent software vendor (ISV) partners, HOOPS AI could become a foundational layer for next-generation design intelligence, much like what NVIDIA’s Omniverse has done for visualization.
It’s also a sign that AI’s next frontier isn’t just text or video—it’s geometry.
Industry Context: A New AI Gold Rush for CAD
Other industry players have hinted at this shift. Siemens has embedded AI-assisted design tools in NX; Autodesk has pushed generative design; and startups are experimenting with ML-driven 3D modeling. But all of these rely on closed systems or limited datasets.
Tech Soft 3D’s approach differs by democratizing access—giving any data scientist or ML engineer the ability to ingest and learn from CAD data without owning the original software.
That could spark a new wave of open innovation across simulation, manufacturing, and industrial AI.
Bridging Design and Data Science
Ultimately, HOOPS AI serves as a bridge between two professional worlds that have operated in isolation: design engineers and data scientists. By speaking both CAD and AI fluently, it promises to unlock value trapped in terabytes of 3D design files sitting idle on enterprise servers.
If successful, this could be one of the most important developments in the evolution of AI for engineering—an enabler for smarter products, faster R&D cycles, and truly data-driven manufacturing.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
