Fortune 100 inventory, redesigned for the people running it.

The system was functional but aging — built around data structures rather than the people operating them. Warehouse staff and plant managers were navigating dense, form-heavy interfaces to manage high-stakes operations across dozens of facilities.

A legacy desktop app the business had outgrown — built around data structures, not the people using them. Eleven user roles, multi-plant access, and FIFO enforcement layered into dense, form-heavy screens. Warehouse operators navigating high-stakes batch actions with no inline validation and no recovery path.

With 11 distinct user roles spanning Normal User through Corporate Accounting, the permissions model was one of the most complex parts of the system. The redesign made it legible at two levels: the role list and the granular permission editor. Plant context was anchored in the sidebar so operators could never act on the wrong facility by mistake, and destructive actions were isolated from save paths to prevent accidental commits.

Status updates and inventory moves were rebuilt to support two mutually exclusive selection modes — separating them visually rather than letting users mix filters eliminates a class of subtle errors the old system surfaced only after submission.

Customer and product records were rebuilt as single-page configurations with inline tables. Add Line No., Add Charge, and Add Pallet keep complex data entry in one flow rather than across multiple pages. FIFO Dates Are Enforced surfaces as a persistent footer warning during order edits — the constraint is visible before commit, not after.

The redesigned system was delivered and put into use. The core shift was one of timing: operators now catch problems before committing them, rather than discovering them after a submit they can’t easily undo.

The constraints that make this domain high-stakes — FIFO enforcement, plant context, the line between an edit and a destructive action — are surfaced at the moment of action instead of buried in the data model. The interface carries the complexity so the people running it don’t have to hold it in their heads.