Proof · Industry · On-prem ML
Categorizer: machine learning for real-world CAD archives.
This page is proof that we apply machine learning to a company’s existing 3D archive — not a storefront.
We built an on-prem system that classifies, compares and searches 3D CAD models in existing company data — without sending that data to third parties.
The situation
Designers and technologists work with thousands of models. Similar designs and duplicates are hard to find, archive knowledge stays with individuals, and repeated work slows quoting and production.
What we built
A system that analyses geometry, dimensions and physical parameters, finds similar models and integrates with existing CAD data management. It runs on-prem.
What it proves
ML on real company data
On-prem operation
Integration with existing archives
Workflow-specific engineering
Technical details
The system analyses 3D CAD models from geometric properties, dimensions and physical parameters. It categorises models, identifies similar designs and duplicates, and supports search across the archive.
It integrates with existing CAD model management systems. No data is sent to third-party services or servers. The system runs on-premise.
Have data where useful relationships are hard to find? Let’s outline an approach.
Contact usHave a concrete problem in testing, data, documentation or AI?
Tell us what is stuck. We can outline an approach before you commit to a project.