Status: RECORDED as standing design input. Stated by Aristide 2026-08-19, during the panel-layout discussion (mockups A/B/C + the Combo + the cases brief). This is not a pick of A, B or C — it is the lens the winner gets designed through, and it carries two firsthand market stories that bear on the whole product, not just the look.
"Most people who grew up in the CAD world have never been exposed to other interfaces. Frankly, they all look the same. Highly technical, a dashboard full of buttons (icons really) and the learning curve is steep. The interfaces definitely do not feel premium nor intuitive — just a bunch of buttons placed next to the last one that was added. No logic. I am coming from the graphic design world, so I have a different perspective. What I am saying is that the users don't know what they don't know, and don't know what they are missing."
What makes this a THESIS and not a taste: it names a structural blind spot in the market's own feedback. The love/hate research reports what operators can compare — and every tool they can compare is an accreted icon dashboard. So the research can tell us what operators hate (and we must fix all of it), but it cannot tell us what a designed interface would win, because no operator has ever been shown one. Evidence has a ceiling; Aristide's outside perspective is the claim about what lies above it. LightBurn is the partial proof already in the data: merely being *cleaner and more visual* than RDWorks took an entire market — and LightBurn is not even design-led, just less accreted.
Design consequences (how the winning mockup gets executed, whichever letter wins):
Jordan had the creators of one of the industry packages at the shop. He asked for a way to modify the file their healer had created. The software engineer's answer: *"why would anybody ever need or want that?"* Jordan had to wrap his way of working around their software; it wasn't flexible at all. He disengaged from the conversation and never adopted the software.
That sentence is where the sale died — and it is the exact defect the research found in auto-nesting (finding f6: output you can't override gets distrusted and hand-edited or abandoned), showing up one stage earlier, in the healer. Recorded as a product principle:
> Nothing cadmaster generates is take-it-or-leave-it. Every stage's output — > healed geometry, nest, tabs, leads, cut order — is first-class editable > material, and an edit re-runs the gates behind it. The operator's judgment > outranks our automation at every stage; the software's job is to make the > automatic answer excellent AND the override effortless, with the checks > re-verifying whatever the human changed.
The pipeline already lives this way (Jordan's hand-moves on the -86/-93 nests, the borderline-chain review flag, grabbable tab/lead markers, operator tolerance overrides). The principle makes it binding on the interface: every mockup element that shows an automatic result must also be the place where the operator changes it.
Everyone in the industry Jordan knows is fighting the same issue: they have access to a healer, but not the flexibility to integrate what the healer created into their own process seamlessly. The healer alone is table stakes; the healer whose OUTPUT drops cleanly into each shop's existing process is the product people actually cannot buy today.
Consequences recorded:
Orthogonal, mostly: A, B and C can each be executed design-led, and the thesis raises the execution bar on all three equally. It does sharpen one line in the cases brief: A's con "weakest demo/sales moment" now weighs more (an interface that LOOKS like every other CAD tool forfeits Aristide's differentiator), and the recommendation "C built in A's order" gains a rider — the dock ships first, but it ships looking designed, not utilitarian.