Find parts by the text written inside the CAD files, find parts by shape, and stop duplicates before they land in your SharePoint library — all running entirely on your own servers.
Protect your data. Protect your designs. No cloud. No phone-home. Works air-gapped.
PremiCAD Search runs beside SharePoint as one small Windows service and adds the search your engineers keep wishing for.
Full-text search over everything written inside the CAD files — part numbers, materials, custom properties, annotation notes — not just SharePoint column values nobody filled in.
Geometry similarity search from a learned shape fingerprint. Works even when naming is inconsistent, wrong, or missing — the model looks at the shape itself.
Check a file against the whole library before uploading it — nothing stored, verdict in under a second — plus a library-wide duplicate report for cleaning up ten years of copies.
The sync can mirror the text it extracts into a plain library column, which SharePoint's own crawler indexes like any metadata. Your search center starts finding CAD parts — nothing installed into your search pipeline.
Numbers measured on a 10,000-part library running on modest shared hardware — the service, SharePoint, and SQL Server on one box.
Overnight initial indexing of a 10k library (~1.5 h in practice); failed files are logged and skipped, never block the crawl.
The footprint is deliberately small: one service beside SharePoint, no farm solution, no schema changes, nothing installed into your search pipeline.
On-premises only, by design — SharePoint Online is not supported and never will be. (SharePoint 2016 and 2019 reached end of Microsoft support in July 2026; Subscription Edition is the on-prem version with a future, and it's what we build and test against.)
Windows Server 2019 or later — and it can be the SharePoint server itself; that is exactly the configuration our numbers were measured on. Budget about 1 GB RAM for the service and ~40 KB of disk per part, plus a local mirror of the CAD files.
Search, indexing, and duplicate checking all run on the CPU. A mid-range GPU is optional and only accelerates the self-learning tier.
STEP, SolidWorks, Inventor, CATIA, NX, Creo, and PRC load through the built-in CAD engine; mesh scan formats (OBJ/STL) are supported too. Assemblies upload as ordinary "Pack and Go" zips.
Read access to the document libraries you want indexed — edit access only if you enable the optional metadata write-back that feeds SharePoint's own search.
No cloud subscription, no external database, no internet connection. One installer, one license file. Uninstall is deleting a folder.
Built for companies with twenty years of proprietary CAD in SharePoint — the kind of data that must never leave the building.
One perpetual license per site, delivered as a single signed file — no per-seat fees, no usage metering, no activation server, no subscription that can switch your library off. Every engineer in the company can search.
An optional yearly maintenance plan — recommended — keeps you current with new releases and support, including the latest CAD format versions as the vendors update them each year. Skip it and the product keeps working exactly as delivered: it's your software. Final pricing is being set together with our beta partners — beta sites run the full product free during the beta and lock in founding pricing afterward.
We are selecting a small number of beta sites with real SharePoint CAD libraries. Beta sites get the full product, direct support, and a say in what gets built next.