Problems first, fields second
We began in 2026 by screening ten industries, then narrowed to five fields: HVAC and mechanical retrofit, construction trades, industrial maintenance, factory automation and warehouse handling. Within them we group what we hear into problem types, such as holding and positioning something heavy or awkward, adapting between parts that don't quite fit, or getting access to a component for service.
Those groups are discovery buckets, not a shortlist. None of them is validated yet, and we say so on our site.
Four gates
Customer problem: is it real, recent, recurring, and costly to the people who do it?
Technical and safety: can it be solved well, and is there any serious hazard we can't resolve? An unresolved hazard stops the process.
Commercial and delivery: will the person who feels the problem be able to buy the solution, and can we make and ship it at a sensible cost?
OEM and IP: does an equipment maker already solve it, or own the space?
Each gate is pass, hold or fail. A problem only reaches engineering once it clears all four.
Then a score, with the uncertainty kept separate
Problems that pass are scored on addressable impact, how often and how widely they recur, how big the gap left by existing solutions is, the purchase case, whether the scope repeats from job to job, delivery economics and access. Unknowns are marked as unknown rather than guessed.
Confidence is recorded separately from the score, two people rate independently, and we check whether the ranking survives small changes to the weights or dropping a single source. If two problems tie, we report a tie instead of forcing a winner.
Why this much process for a tool
Because the expensive mistake isn't a bad prototype. It's a well-engineered tool for a problem nobody pays to solve. The gates are how we avoid it.