CAD-to-mesh transfer auditing for Altair workflows

Every time a geometry file moves from a CAD tool into Hypermesh — especially from ANSA, NX, or CATIA — something gets lost or corrupted. Users report 'a lot of data loss when importing from ANSA to Hypermesh' and geometry imports taking significantly longer in newer Hypermesh versions than in old ones. Engineers currently discover these translation errors only after they've already spent time meshing, or worse, after a solver run fails in a way that traces back to a geometry artifact nobody caught at import.

This gap persists because the CAD vendors and Altair each own one side of the interface and have limited incentive to deeply audit what happens in the handoff — the data leaves one system and arrives in another, and neither side is responsible for what the engineer finds in the middle. The people who feel this most acutely are meshing engineers who do the same import-check-cleanup cycle dozens of times per project and have developed ad hoc personal checklists to catch the issues they've been burned by before.

Nothing today gives an engineer a structured diff between what was in the source geometry file and what landed in Hypermesh — which surfaces disappeared, which small features were dropped, which tolerances changed. Engineers do this manually by eye, which is slow and misses subtle issues that only show up in solver results. The workflow is inherently repetitive: every project, every geometry revision, every import. That's not a one-time problem to solve; it's a recurring tax on every simulation cycle.

The business case is concrete: a missed geometry translation error that propagates into a structural or thermal simulation can invalidate weeks of compute time and engineering interpretation. The engineer who catches it at import saves the project; the one who doesn't catches it at report review or, worse, in physical testing.

What to build

Build a desktop utility that sits between a CAD export and a Hypermesh import, parses both the source file and the resulting HM model, and produces a structured diff report flagging missing surfaces, tolerance changes, dropped small features, and entity count mismatches — exportable as a PDF the engineer can attach to their project record.

Where to start

Start with ANSA-to-Hypermesh specifically — it's the exact import path users named in complaints, it's common in European automotive supply chains, and it has the most documented frustration — which makes the first customer conversation easy to anchor.

The hard part

The utility needs to parse a wide range of CAD formats (STEP, JT, CATIA native, ANSA .ansa) with enough fidelity to produce meaningful diffs — that's a non-trivial geometry kernel problem, and getting it wrong for even one common format kills trust immediately.

How it makes money

Per-seat annual license, priced as a productivity tool rather than enterprise software — roughly $500–$1,200/year per engineer — with volume discounts for teams of 10+.