Restaurant Digital Twin
A five-store restaurant group with sales, recipes, ingredients, purchasing, inventory, cost and profit linked into one model. The ingredient twin shows what one kilogram of rice was plated into, the revenue of those dishes, the profit allocated to rice by an explicit rule, and what the rice cost.
- stores, 90 days of transactions
- 5
- dishes / ingredients with recipes
- 60 / 120
- suppliers with price history
- 15
- step guided demo, anomaly to approved PO
- 9
The challenge
A margin drop at one store hides inside thousands of transactions. Was it a supplier price change, over-portioning, waste, a promotion or labour? Finance sees it a month later.
What we built
- Seeded, reconciled dataset: transactions, perpetual weighted-average inventory, POs, waste, labour, opex and budget
- Calculation engine independent of the UI: P&L, dish metrics, ingredient attribution by cost, quantity or equal share
- Forecasting and replenishment with approvals and an audit trail
- What-if and event simulators for suppliers, sales and operational incidents
- Bilingual English / 简体中文 interface
Outcome in the twin
The guided demo traces a store’s margin anomaly to a local rice supplier and over-recipe usage, then ends with an approved purchase order and its verified inventory effect — all reconciled to the cent by 22 tests.
Path to production
Connect POS, accounting and supplier invoices; the engine already separates data from calculation, so real feeds replace the seeded generator.
Technology
About this build
Fictional group; all data seeded and synthetic; no POS or supplier system connected.
Warehouse Twin
Digital Energy Factory
Have a system you want to see, question or rehearse?
Tell us about the site, the process and the decision. We will come back with what a twin could answer, which data it needs, and how fast a prototype can be in front of your team.