Honest Reflection
The Assembly Item backend compromise is the trade I'd revisit if I had the chance. The compromised data model means certain future features particularly multi-level BOMs at scale and assemblies-of-assemblies will be harder to build than they would have been with the full item-type approach. The call was right for the February timeline; the cost is real for the next 12-18 months of manufacturing roadmap. If I had this back, I'd push harder for the cleaner backend architecture even if it meant a partial February delivery but still ship the highest-frequency workflows on the right architecture, defer the edge cases, and avoid the compounding tech debt. The lesson: 0-to-1 under compression compounds tech debt at a rate you don't feel until the next product cycle. I'd also build a stronger feedback loop with the customers who paused their competitor exits. Right now I know they stalled their decisions but I don't yet know which specific design decisions tipped them. That insight would shape the next generation of manufacturing work, and I didn't structure the research to capture it.
