loh.ai

Automated surface inspection.

loh.ai checks the 3D scans you already take for surface defects, grades each one and writes the inspection report in seconds.

Join our waitlist

In partnership with:

  • NVIDIA Inception

  • Microsoft for Startups

  • University of Oxford Incubator Cohort

  • LSE Generate

Drag to turn ·

Start with the scan you already take

Export the STL or mesh from your 3D scanner as you do now. The example here is a machined plate, 60 × 40 mm, and every number in this walkthrough comes from its scan.

D1D2D3D4D5D660.040.0

Machined plate, 60 × 40 mm

15,000 points · 0.40 mm spacing · raw scan

Every point measured against the part itself

loh.ai fits a surface to the part and measures how far each scan point sits from it. Points past the threshold are flagged at the same sensitivity on every part and every shift, so the result doesn't depend on who is inspecting or how long they've been at it.

D1D2D3D4D5D660.040.0

Machined plate, 60 × 40 mm

Pit −47 µm+47 µm Bump

Defects found and ranked in seconds

Points that stray are grouped into defects. Each one is classed by type and graded on its area and how far it deviates, so your team can start with the critical ones.

D1D2D3D4D5D660.040.0

Machined plate, 60 × 40 mm

4 critical1 moderate1 minor

The inspection report is written for you

Each scan produces a report giving every defect's location, type, size, depth and grade, with a plain-English summary of the result. You don't need a metrology specialist to read it. If a finding needs explaining, an assistant answers questions using that scan's measurements.

loh.ai
Report no.SAMPLE-001
Issued23 Oct 2026
RevisionA

Surface inspection report

Automated 3D scan analysis

1Part and scan

PartMachined Plate
Source filetest_plate.stl
Envelope60.0 × 40.0 × 0.9 mm
Points analysed15,000
Median point spacing0.403 mm
Surface variation (1σ)9.3 µm
Detection sensitivity8.0

2Result

FindingsCriticalModerateMinor
6411

3Summary

Six surface defects found across 15,000 measured points. Four are critical, one moderate and one minor.

The critical findings are two pits and two bumps, each covering at least 10 mm². The largest is pit D1 at 26.8 mm², 58 µm deep.

Pit D6 covers 1.6 mm² across 10 points at 2.9σ, close to the scan's noise floor. Re-scan that area before acting on it.

4Findings

IDTypeGradeArea (mm²)Max. dev. (µm)σx (mm)y (mm)
D1PitCritical26.76−586.224.825.1
D2BumpCritical22.38+9410.115.19.9
D3BumpCritical15.24+10611.440.130.1
D4PitCritical10.22−12513.49.932.1
D5BumpModerate8.11+10911.749.911.9
D6PitMinor1.62−272.928.035.0

Max. dev. is measured from the fitted surface: negative for material missing (pit), positive for material added (bump). Locations are finding centres from the lower-left corner.

loh.ai · Surface inspection report · SAMPLE-001Page 1 of 2
loh.aiSurface inspection report · Machined plate

5Defect map

D1D2D3D4D5D660.040.0
Figure 1. Plan view of the inspected surface with findings marked. Dimensions in mm.

6Method

Each scan point is compared with a surface fitted to the part. Points that deviate beyond the detection threshold are clustered into findings, classed as a pit or a bump by the direction of deviation, and graded on the area they cover and their maximum deviation.

Prepared byloh.ai automated inspectionReviewed byDate
loh.ai · Surface inspection report · SAMPLE-001Page 2 of 2

Written automatically from the measurements in this scan.

Works with the scans you already take

Export the mesh the way you do today. The results come back with the report attached.

Specification

loh.ai inspection

Input
STL or mesh export from any 3D scanner
Scan size
Any
Processing time
Seconds per scan
Defect types
Any defect type
Grading
Critical, moderate or minor, by area and deviation
Output
Marked-up defect map and written inspection report
Follow-up questions
Answered from the scan's own measurements
Integration
None needed to evaluate
Your data
Stays yours

See the report for one of your own parts

Bring a scan to a 30-minute call.