This case study explores a fictional relationship, intended to help you understand our brand-new product, HOOPS AI.

This resource is intended to showcase a potential use case of this toolkit to provide context on its capabilities and potential. To explore Tech Soft 3D's real relationships with our partners, we encourage you to check out our partner success stories for 50+ examples.

The Challenge:

Ziemens is a leading digital manufacturing marketplace that connects engineers and designers with a global network of qualified manufacturers. The company provides on-demand production services for custom parts using advanced manufacturing technologies, including CNC machining, 3D printing, injection molding, and sheet metal fabrication. Users can upload CAD files, receive instant pricing and lead time estimates, and place orders directly online. Ziemens’ web platform simplifies the manufacturing process by streamlining supplier selection, improving supply chain visibility, and ensuring high-quality production at scale. It serves a wide range of industries, including aerospace, automotive, medical, and industrial equipment.

For Ziemens, juggling the demands of speed and accuracy is a key balancing act. Their customers use their tools to automatically interpret CAD data from a variety of sources, determine manufacturing feasibility, and generate cost-optimized production quotes in minutes. But as models and assemblies became more complex, the Ziemens team faced an enormous challenge: how to provide more accurate quotes automatically to their users.

While AI tools have streamlined work in many other industries, the unique and complex nature of CAD data made engineering software notably lag behind. Ziemens' new AI Assemble tool provides its users with an unparalleled ability to automatically process, qualify, and leverage CAD data into actionable CAM insights.

The Unique Challenge of Using AI in 3D Engineering

HOOPS AI is the newest toolkit from Tech Soft 3D that provides organizations with the fastest path from CAD to machine learning models. Designed to handle the entire pipeline, from data ingestion, cleaning, encoding, and preparation all the way to ensuring reproducible runs and reliable versioning, this toolkit helps organizations build a “factory” for experimenting with CAD machine-learning tools. Built on the reliable foundation of HOOPS Exchange, Tech Soft 3D’s CAD file access toolkit, HOOPS AI comes with visualization and interpretation tools specialized for working with CAD data.

For partners like Ziemens, HOOPS AI enables the ability to experiment rapidly, supporting the unique workflows that developing AI tools requires.

“We quickly realized that building machine learning workflows with 3D data is nothing like building traditional CAD applications. It’s not just about hiring developers and coding, it’s about experimenting, learning, and improving over hundreds or thousands of cycles.”

Lukas Bauer, Product Manager at Ziemens.

Building Machine Learning Workflows Around CAD Data

The biggest barrier for Ziemens was not creating the AI models, but everything that had to come before the models could be trained. Turning CAD files into data usable for machine learning requires a massive amount of preprocessing, including:

• Ingesting thousands of CAD files in different formats

• Geometry, topology, assembly hierarchies, PMI, tolerances, and material properties have to be extracted and cleaned.

• Relationships between parts and features must be preserved

• All data had to be encoded numerically so ML frameworks could understand it.

Historically, these tasks consumed 70–80% of any CAD-based ML project, slowing innovation and adding risk.

HOOPS AI slashes this timeframe by providing a unified, scalable pipeline. Powered by HOOPS Exchange and built to support the most time-consuming aspects of this process, the toolkit offers:

• Automatic ingestion and conversion of multi-format CAD files

• Built-in repair, normalization, and assembly resolution

• Feature and part classification with embedded PMI extraction

• Graph-based and geometric encodings ready for machine learning

• Parallelized pipelines for hundreds or thousands of experiments

• Data storage, versioning, and visualization tools for traceable, reproducible results

“With HOOPS AI, we were able to set up a full training and evaluation pipeline in weeks instead of months. It let our data scientists focus on improving the models rather than spending all their time on infrastructure.”

Lukas Bauer, Product Manager at Ziemens.

HOOPS AI’s built-in visualization and orchestration tools helped Ziemens run hundreds of experiments in parallel, measure model performance, and continuously refine results. “The experimentation system was a game-changer,” said Lukas. “We could test new architectures, adjust hyperparameters, and compare results instantly. It gave us a ‘factory of experiments’ that kept improving itself. Our first successful run happened in hours instead of weeks.”

The Results: From Data Overload to Unparalleled CAM Insights

With HOOPS AI, Ziemens developed new capabilities to predict the best production process based on each part’s geometry, material, and PMI. From automatically recognizing where tighter tolerances are needed to flagging areas where different manufacturing methods may be needed, Ziemens’ AI Assemble software empowers CAM companies to make smarter decisions from their CAD data.

“The key isn’t just recognizing a hole or a face, it’s understanding why that hole exists, what it connects to, and how precise it needs to be. That’s what HOOPS AI helps us do. It turns raw CAD data into manufacturing intelligence.”

Lukas Bauer, Product Manager at Ziemens.

The power of HOOPS AI enabled Ziemens to drastically reduce their development time and costs. “We estimated that by using HOOPS AI, we cut 50% off our time-to-market, going from nine months to just over four,” Lucas explained. “The HOOPS AI SDK helped us do more with less. We were able to concentrate a smaller group of engineers on quoting and get more out of our existing team than we thought possible.”

This rapid iteration process has already produced results. The latest version of Ziemens’ AI Assemble now includes an intelligent design-assist feature that automatically detects potential manufacturing issues and proposes geometry adjustments.

The Ziemens team continues to expand its use of HOOPS AI, exploring applications in generative design. As Lukas summed up, “HOOPS AI gave us the foundation to move faster, take more risks, and innovate with confidence.”

At Tech Soft 3D, we’re proud to support Ziemens and other partners who are building the future of intelligent engineering software.

This case study explores a fictional relationship, intended to help you understand a brand-new product. To explore Tech Soft 3D's relationships with our partners, we encourage you to check out our partner success stories for 50+ examples of Tech Soft 3D's work with real companies.