A new drawing package lands from the M&E consultant. Forty PDFs, named things like 2026-114-M-201_C.pdf. You open one. The title block says Rev C. A revision table above it lists all the previous revisions, and there's a job number, a drawing number and a sheet number all sitting within a few millimetres of each other. The author is listed as "Harper Building Services", but in your project they're set up as "Harper Limited Ltd".

You'd make sense of all that in a few seconds. Now do it forty times, then again next week when the next package arrives.

That's the job document software is meant to take off your hands. How well it does it comes down to one choice: does it just read the drawing, or does it understand it?

What our rivals use (OCR)

OCR (optical character recognition) turns what's on a page into text, and it's very good at it. What it can't do is understand what that text means. To OCR, "Rev C", "Kitchen", "1:50" and "Do not scale" are all just words on a page.

To make it useful, OCR systems rely on templates that say where to look: drawing number in this box, revision in that one. That's fine if every drawing comes from one architect with one title block. On a real job you've got the architect, structural engineer, M&E consultant and a handful of specialist subcontractors, all with different layouts. The template breaks, and someone's back to typing it in.

How Drawli does it

Drawli uses AI that reads the whole drawing the way you would. It finds the title block wherever it sits, tells a drawing number from a job number, and recognises an elevation from what's on the sheet. From every upload it picks out the document type, title, number, revision, author, location and trade.

But reading the drawing is only half of it. Drawli also checks what it finds against your project.

It knows who's who

Drawli cross references every drawing with the companies already in your project. If the title block says "Harper Building Services" and your project has "Harper Limited Ltd", it recognises they're the same firm and matches them, rather than creating a duplicate you have to tidy up later.

It puts revisions in the right place

Drawli reads the revision table, not just the revision box, and checks it against the drawings you already hold. Upload Rev C and it's filed as the current version with A and B marked as superseded. Upload a Rev B that turned up late and it slots in behind C, instead of overwriting the drawing everyone should be working from.

It reads drawings from abroad

Working with a designer or manufacturer overseas? Drawli understands title blocks in other languages and translates the details, so a German "Grundriss Erdgeschoss" is filed as a ground floor plan like everything else.

Why it matters

Filtering, revision tracking and knowing who's got the latest set all depend on the details going in being right. OCR gives you text and leaves you to sort it out. Drawli gives you a drawing that's already filed, matched and in the right place.

Try it

Start a 2 week free trial at www.drawli.cloud and upload a few drawings from different consultants. See how it gets on.