A generalisation memo answers one question after each engagement: what did we build here that the product should absorb, what was genuinely bespoke, and what should we never do again? It is the mechanism that converts services work into compounding product value, and roughly a third of a frontier-lab FDE’s time is nominally spent on this kind of feedback.
Worked example: an FDE who ships six deployments and turns two recurring patterns into product capability reaches staff level; one who ships ten excellent bespoke deployments and generalises nothing does not. Gotcha: if a deployment organisation has no such artefact anywhere, it has quietly become a consultancy — which is the clearest early symptom of the services trap.