Reads drawings like an engineer.
One API-first extraction engine. Send a drawing, ask for the depth you need, and get clean, typed JSON back in seconds — no model training, no templates.
Pick the depth, per drawing.
Three extraction SKUs, mixable in a single call. A page costs the same whatever you ask for.
Title-block metadata
Drawing ID, designation, material, weight, tolerances, projection, BOM and notes — straight from the title block.
- drawing_id
- material
- weight
- projection
- BOM
Geometry & features
Dimensions, fits, threads, bores, chamfers, GD&T, radii and surface roughness — each with confidence and x/y/bbox coordinates.
- ⌀ dimensions
- H7 / G6 fits
- M-threads
- GD&T
- Ra
Manufacturability
Primary & secondary processes (DIN 8580) and volume and bounding-box estimates for quoting.
- DIN 8580
- volume est.
- bbox
Send what you already have.
Vector or raster, clean or scanned, single- or multi-page (as PDF). ISO and ASME / ANSI, across six languages.
From PDF to JSON in three lines.
- REST API, webhooks & an official Python client
- Synchronous or async callbacks for batch archives
- Every field carries confidence + source coordinates
- Output in JSON, CSV, XML — or your custom schema
Shaped to your workflow.
Results in your format.
No templates, no parsers. We match your ERP/PLM/MES schema and return data in your field names and structure — so it lands ready to use, not ready to map.
Anonymize & restyle.
Strip identifying marks and re-style incoming drawings to your own template — useful for RFQs, supplier sharing and archival standardization. Anonymizing is in the API; restyling to your template is on a contract.
Send a PDF. Get structured JSON.
Test Werk24 on your own drawing in 60 seconds — no integration required.