Lightspeed’s retail agent brings a useful question: buy more, or move the stock?

A shop’s last jacket can prompt a stock transfer as well as a purchase. That is the kind of decision behind Lightspeed’s 15 October announcement: its retail AI can use sales, stock and purchasing records to prepare actions, including purchase orders and transfers between locations.[1]

The new actions are being introduced through an opt-in beta for selected customers. Lightspeed says the retailer sees a preview, approves a change before it happens and gets an audit record. This is an analysis of the announced capabilities and current inventory documentation; we have not used the beta.[1]

One business, two very different shelves

Consider an invented retailer with two branches selling the same jacket in the same size. For the next fortnight, North expects demand for 24 units and South for six. There are no outstanding orders or reserved units, and both branches can share stock before the forecast period begins. These are deliberately simple assumptions, not a Lightspeed customer result.

North is short by 16 units, but the business holds 38 against combined demand of 30. Moving 16 from South fills North’s gap and leaves eight units after both branches’ assumed sales. Buying 16 new units instead would raise opening stock to 54 and leave 24 afterwards. The transfer avoids an additional purchase; it does not conjure up extra sales.

Lightspeed’s existing X-Series documentation already recognises this distinction. In its distribution model, the suggested order calculation considers stock across the store network and can return zero when enough could be relocated. A separate warehouse view includes stock held centrally; direct ordering focuses on an individual outlet. Those are different ways of asking how much the business needs to buy.[2]

Demand is more than the number that sold

The forecast also needs to understand an empty shelf. Lightspeed’s documentation uses in-stock days when estimating a product’s daily sales rate and offers an option to factor estimated missed sales into demand. Its seasonal calculation uses the corresponding prior-year period for products with recurring seasonal peaks. A branch that sold little because it had no stock should not automatically look like a branch with little demand.[3]

The jacket example leaves no buffer at North. A retailer expecting uncertain demand might retain extra stock, order more or transfer in stages. Transport cost and arrival time also belong in the decision: an inexpensive internal transfer that arrives after the customer has gone cannot solve today’s shortage. The useful comparison is between feasible ways to meet demand, including the inventory left afterwards.

The transfer ends when the goods arrive

In the documented X-Series process, an accepted stock request becomes an open transfer. Stock leaves the source outlet’s recorded inventory when the transfer is marked as sent. After physical arrival, the destination records the quantity it actually received. A proposed movement, goods in transit and goods available on the shelf are separate stages.[4]

The beta’s commercial promise is to shorten the path from spotting the imbalance to preparing the right transaction. In this example, the decisive output would be a proposal to send 16 of the correct jacket variant from South to North, with an arrival time that makes the transfer useful—not just a message saying North is running low.

Report an error or suggest a correction