Purchasing

AI purchasing agent for a café · pilot

In progress

A forecast every morning: what runs out within 7 days, what is sitting in excess. A draft order per supplier in that supplier's packaging, a human says "ok", the order goes out by email or WhatsApp, the delivery is booked in the stock system on the day it arrives. The goal: no stop-lists, no excess stock, the warehouse kept at a target number of days.

Numbers will follow after launch.

Client

Café, Limassol, Cyprus.

The client is kept anonymous.

How it was

Stock was counted and supplier orders were put together by hand.

What we did

1. Every morning: a forecast of what runs out within 7 days and what is in excess.
2. A draft order per supplier, in that supplier's packaging.
3. A person approves it in one tap; the order goes out by email or WhatsApp.
4. The delivery is booked in the stock system on the day it arrives.

Result

• 7 daysforecast ahead, every morning
• 1 tapto approve each order
• Same daydelivery booked in stock

Stack

Claude API · Make · Airtable · stock system