The Honest AI Question for a Made-to-Order Platform

Founders almost always want to talk about where AI goes in the product. With Zammit, the most useful thing I can offer is a disciplined answer rather than a hopeful one. This was a platform built on the DriveWorks configurator engine, where the value came from deterministic, rule-bound configuration: a contractor or retail professional sets dimensions, shelf count, edge profiles and placement, and the output has to be fabricable on a laser-cut line every single time. In a domain like that, the bar for any AI is whether it respects those manufacturing constraints. The honest place to look isn't replacing the configurator — it's reducing the cognitive load of self-serve. The real friction we designed against was non-experts configuring complex made-to-order products without CAD knowledge and without a sales intermediary. That's exactly the kind of guidance, interpretation and 'did you mean this configuration' assistance an AI strategy should be aimed at — never at generating specs the factory can't cut. The founder takeaway: anchor AI ambition to the actual job the product removed a human from doing, and let the engine's hard constraints, not the hype cycle, set the boundary.