Constraint and tolerance audit tool for simulation-bound CAD

The moment this problem surfaces isn't during modeling — it's when an engineer hands off geometry to a simulation team and the mesh fails, or when optimization loops break because the geometry has hardcoded values that can't be parameterized. Users say 'making parametrizing a geometry for optimization is not simple' and that the 'step by step slow process' of modifying a CAD model and propagating changes makes design-of-experiments workflows nearly impossible in direct-edit environments. A related complaint: 'the maximum fillet radius allowed in mm is 0.1 mm, requiring a change of the unit system for smaller radii' — this kind of hidden constraint causes silent errors downstream that engineers only discover after a simulation run fails.

This gap persists because the people who feel the pain most — simulation engineers and CAE analysts — are not the people who built or purchased the CAD tool. The CAD tool was bought by the design team; the simulation team inherited it and has to live with its constraints. Neither group has the leverage or the budget authority to get the modeling tool fixed, so the simulation engineers work around it manually — re-meshing, hardcoding values, writing notes in spreadsheets.

What's missing is a pre-simulation geometry audit: something that reads a CAD file before it enters the meshing or optimization pipeline and flags every hardcoded dimension, every geometric constraint that would prevent parameterization, every unit-system anomaly (like the 0.1mm fillet floor), and every feature that historically causes meshing failures — then outputs a structured report an engineer can act on before the simulation job is queued. This doesn't require changing the CAD tool at all; it sits between the CAD output and the simulation input, which is exactly why an independent product can own it.

Simulation compute time is expensive — cloud HPC costs real money per hour — so catching a geometry problem before the job runs, rather than after, has a concrete and immediate dollar value. That's the business. The need recurs every time a new geometry version is passed to simulation, which in active development programs happens multiple times per week.

What to build

Build a CLI and web UI that ingests STEP/IGES files, statically analyzes them for parameterization blockers (hardcoded radii, unit-system anomalies, non-manifold edges, features incompatible with common mesh generators), and produces a prioritized punch list engineers resolve before submitting to simulation — with direct integration into Ansys Workbench project pipelines.

Where to start

Enter through Ansys Workbench users specifically, where the SpaceClaim geometry handoff to simulation is a documented and complained-about friction point — the integration path is known, the user complaint is public, and the buyer (simulation team lead) is different from the CAD tool owner, giving you a sympathetic internal champion who didn't choose the broken tool.

The hard part

The hardest early challenge is building a detection library for 'parameterization blockers' that is specific enough to be actionable and not generate so many false positives that engineers ignore the report — getting that signal-to-noise ratio right requires deep access to real simulation failure logs, which means the first customers have to be unusually cooperative.

How it makes money

Usage-based pricing per geometry audit run, with a monthly cap for teams running high-frequency optimization loops — simulation teams already think in per-run costs, so this pricing model matches how they already budget compute.