Three years running document control on a structural steel shop floor taught me exactly where the manual work piles up. Now I build the AI and automation that gets rid of it — for shops that don't have time for a software project that doesn't fit how they actually work.
Every detailer packages their drawings differently — different folders, different naming, different everything. Somebody still has to open each PDF, check it against the title block, rename it, sort the bolt lists and CNC files, and get it into production. That's real hours every week, and it doesn't get any cheaper as you take on more work.
A ZIP filename can say anything. A folder name can say anything. The drawing itself — the title block — is the legal document of record. So that's the only thing the system trusts.
There's no dashboard to learn and no training session. Tell it what you need in plain English — process this transmittal, check this job's status — and it runs the pipeline and tells you what it found, including anything it's not sure about.
Every detailer uses a different folder structure, a different title block layout, a different page size. A one-time setup — about 15 to 30 minutes — teaches the system that detailer's format. After that, every package from them processes on its own, for as long as you keep using it.
Every behavior below exists because something like it went wrong once, by hand, and cost real time to track down.
Normalizing packages — classifying, renaming, verifying against the title block — is running today. Next is automating the import into your production system with checks against the source drawings, then automating transmittals after that. Each phase cuts out another manual step.
No pricing conversation. No slide deck. Send a package from your own detailer and see exactly what comes out the other end.