Make know-how reusable

“lack of established digital workflow”

Borg et al., Advanced Engineering Informatics, 2024

Experience should change the system

A company’s most valuable design knowledge often lives in decisions people make without writing them down: which constraint can move, which machine tolerates a variation, why a layout rule exists, and which trade-off a client accepted. When that knowledge stays in a person or a one-off file, the next team must rediscover it. Turnover makes the loss visible, but the cost is present even while everyone remains employed.

The priority should be to capture know-how in the most reusable form that is practical for the work. Sometimes that is a searchable decision record. Often it is a parametric model, validation rule, shared schema, tested toolpath, or explicit mapping between objectives and variables. The point is to preserve reasoning that can be applied to the next problem, rather than archiving every output.

What the papers tell us

In our paper on generative design for papermill layouts, the team found that adoption was constrained by a “lack of established digital workflow” and limited in-house generative design experience. Those are operational problems as much as software problems. A generator has value when the client can maintain its rules and trust the process that produced its options.

Our later Knowledge Augmented Generalizer Specializer paper tackles an earlier stage of the problem. It describes a method for “problem formulation, hierarchical decomposition, and solution generation.” Its subsea layout case study is a research demonstration, not evidence that any one company has solved knowledge retention. My wider takeaway is that design systems should make assumptions, parameters and relationships explicit enough to inspect, challenge and reuse.

Shift the effort

I would rather spend the next hour improving a team’s ability to solve a class of problems than re-creating the same answer for a new brief. That shift needs ownership, documentation and feedback from the people doing the work. Good automation is a living capability: when practice changes, the system can change with it.