Upload STEP, IGES, DXF, PDF or even a phone photo. AI extracts dimensions, ISO tolerances, Ra finishes and GD&T symbology error-free thanks to the 4-layer guardrail (extractor → validator → reasoner → safety score) that blocks LLM hallucinations.
STEP, IGES, DXF, PDF, JPG/PNG. No manual conversion: every format is normalized to an internal B-Rep + rendered bitmap, then fed to the AI engine.
Extracted L/W/H/Ø appear as colored bounding boxes on the drawing. Click a dimension → real-time highlight. Manual edit → automatic quote recompute.
Detects and classifies through holes, metric threads, grooves, pockets, chamfers, fillets, slots, counterbores, countersinks. Each feature is tagged with severity (info/warn/alert) to anticipate machinability issues.
Layer 1 (extractor) extracts measurable data only. Layer 2 (validator) checks ISO coherence. Layer 3 (reasoner) reasons with strict anti-fabrication prompts. Layer 4 (safety) computes a 0-100 safety score and auto-blocks output below 70.
“We used to redraw customer PDFs in CAM by hand. Now the AI extracts 12-15 dimensions in 8 seconds. ~20 min saved per quote.”
Answers to the questions we hear most often from workshop owners and CAM programmers.