A design-service network spread across several cities had strong design capability and no shared standard for how work arrived or got checked. Requirements came in whatever shape the sender chose, and whether a plan passed depended on which reviewer opened it. Over roughly two quarters we built the intake, generation, and validation layer that gave every location one way of working.
2 quartersmulti-phase rollout
A regional design operation needed consistent output across multiple cities while keeping costs predictable. Demand grew faster than manual drafting capacity, and turnaround pressure started showing up in review quality.
First we mapped how intake and review actually worked, city by city. Then one shared data model for requests, constraints, and acceptance criteria — the thing every location had been interpreting its own way.
From that model: structured intake forms, automated CAD starter generation, and a validation engine that encodes their own review standards — room dimension minimums, light and ventilation per room, wet areas stacked between floors. Checks that used to live in individual reviewers' heads now score a plan before anyone opens it.
Then runbooks, ownership checkpoints, and weekly reliability reviews — so the workflow held as volume grew, instead of buckling under it.