Inventory & cost intelligence

The challenge

A multi-outlet kitchen was closing the month by hand. Paper counts, invoices keyed one line at a time, and no clear picture of what anything actually cost. Seven distributors each sent their data in a different format.

What we built

A hosted platform that runs the whole cycle. Staff count on an iPad with a tap-to-count interface, each working their own area at the same time, with every edit tracked. Invoices arrive two ways: clean vendor files are parsed instantly, and anything scanned or photographed is read automatically and held at a reconciliation gate before it posts. A Cost Watch view normalizes every item to a true per-unit price and flags real increases while ignoring the false spikes that promo pricing creates. Month-end finance export drops out as a finished workbook.

Under the hood

Next.js, Supabase (Postgres and auth), and Vercel, with automatic reading for scanned and photographed invoices. A thirteen-table schema, role-based access, and a full audit trail. Seven vendor formats parsed and validated, 146 duplicate item records cleaned up, and over 800 price points loaded and verified.

The outcome

Month-end went from a manual slog to a guided process, with live cost visibility the operation never had before. That visibility is the foundation for catching the money leaks that used to go unnoticed.

Dashboard showing live inventory value, suggested order, and latest count status — The operator's home screen — live inventory value, suggested order, and where the month-end count stands.

Tap-to-count screen on an iPad, listing kitchen items with a quantity field for each — Tap-to-count on an iPad. Staff each claim an area and count in parallel, every entry saved as they go.

Invoice review screen flagging a mismatch between printed total and line items — Every invoice is reconciled before it posts. A $4.50 mismatch is caught and held at the gate — with the price jump flagged — instead of slipping through.

Cost Watch table normalizing per-unit prices and flagging the biggest movers with trend lines — Cost Watch normalizes every item to a true per-unit price and puts the real movers first, ignoring the false spikes promo pricing creates.

Month-end reports screen valuing inventory by area with an export button — Month-end valuation by area, exported to a finished finance workbook in one click.

Next.js · Supabase · Vercel · automatic invoice reading

← Back to work

Inventory & cost intelligence

The challenge

What we built

Under the hood

The outcome

Want something like this?