The Most Expensive Product Mistakes Happen Before Engineering Begins
Why the brief — not the BOM, not the tooling — is where hardware projects usually win or lose.

Insights
Five pillars of perspective drawn from real product work — on strategy, engineering, manufacturing, AI, and what it actually takes to bring hardware to market.
01 / Pillar
What should companies build — and why?
The decision of what to build is where the most expensive mistakes are made. By the time engineering starts, most of the cost is already locked in.
Why the brief — not the BOM, not the tooling — is where hardware projects usually win or lose.
The hardest discipline in hardware is knowing when to stop. What the signals look like before the money is spent.
The better question is what problem is worth solving — and for whom the answer is worth paying for.
02 / Pillar
Lessons from solving difficult physical-product problems.
A beautiful concept and a manufacturable product are not the same object. The work is making them the same object.
When form wins the review but loses the field — and how engineering judgement has to push back on the render.
A prototype that proves the idea is not a prototype that proves the product. The gap between the two is where most projects stall.
The hardest limits — assembly sequence, service access, human hands — rarely appear in a requirements document.
03 / Pillar
What designers and founders misunderstand about bringing hardware to scale.
A design that can’t be built repeatedly is a prototype that got lost. Scale is its own design problem.
A factory that just says yes is a liability. The best manufacturing relationships are arguments — productive ones.
Draft, tolerance, flash, gate placement, assembly order — the realities that turn a clean model into a yield problem.
Stack-up is not a detail. It is the difference between a product that snaps together and one that never will.
04 / Pillar
How AI is changing product development — and creating new physical-product opportunities.
AI accelerates the easy part of hardware. The hard part is still atoms, tolerances, and the world.
Idea volume was never the bottleneck. Selection, feasibility and making it real — that’s where the work is.
Where AI runs out of leverage is exactly where physical-product expertise still decides who wins.
Generative tools accelerate exploration and analysis. They don’t replace the judgement that ships a product.
05 / Pillar
Why technically excellent products still fail — and what to consider earlier.
A product that works is not a product that sells. The conditions for market success are built in long before launch.
Performance isn’t enough. Channel, positioning and timing decide whether the right product reaches the right person.
By the time you’re ready to launch, the decisions that determine whether launch works are already made.
How a product is sold shapes what it has to be. Treating channel as an afterthought is a late-stage tax.