Geometry prep and cleanup for simulation engineers

The moment that drives someone to look for this is when they've just finished a topology optimization run, they have a promising structural result, and they realize they can't turn that mesh output into a clean, editable CAD surface without leaving the tool entirely — or when they need to clip a section across a 40-part assembly and there's simply no way to do it. This gap persists because the simulation vendors (Altair, Ansys) have a structural incentive problem: their revenue and brand equity live in the solver, not in geometry editing. Every dollar they spend on CAD features competes with investment in solver accuracy, which is what wins enterprise contracts. The buyer is a simulation manager who cares about solver licensing; the daily user is an engineer who wastes two hours per day on geometry prep. Nobody at the buying level is screaming about it.

What's broken in practice: users report that TO results can't be converted to CAD surfaces effectively, that there are no tools for point import from text to generate truss geometries, that section clipping and measurement are missing, and that when a CAD file has multiple components the tool often drops surfaces on import. These aren't bugs — they're absence-of-feature problems that a focused geometry product could own entirely.

This is a business rather than a feature because the need recurs on every simulation project. An engineer iterates geometry dozens of times per model before the mesh is even set up. Without a clean intermediate layer, they're licensing a full CAD seat (expensive), or they're manually rebuilding geometry in a separate environment every cycle (slow and error-prone). A dedicated tool that sits between CAD and solver — handling cleanup, section editing, parameterization, and TO-to-surface conversion — justifies its own seat license precisely because the alternative costs are quantifiable per project.

What to build

Build a standalone geometry editor that imports simulation mesh or TO output and exports clean, parametric STEP/IGES geometry — including point-cloud-to-surface reconstruction, section clipping with measurements, and a coordinate/expression-based point importer for structural geometry like trusses.

Where to start

Start with truss and frame structure generation from coordinate input — this is a narrow, well-defined workflow (SpaceClaim users literally asked for point import from text code) where a scriptable tool wins immediately over a GUI-only CAD package, and structural engineers will pay to avoid switching contexts for a task they do every week.

The hard part

The hardest early problem is TO-to-surface reconstruction quality — users have specifically called this out as ineffective in existing tools, and building a robust enough implicit-to-NURBS pipeline that engineers trust the output enough to stop manually rebuilding is a non-trivial computational geometry problem with a high correctness bar.

How it makes money

Annual per-seat license, priced below a full CAD seat but above a mesh utility — roughly $800–1,500/seat/year, with volume discounts for simulation teams of 5+ engineers.