Singapore Mall AI Fridges and a Dutch Weight-Sensing Fridge: Two Purchase Records
The Singapore customer installed multiple AI vending machines in a shopping mall for snacks and beverages. The Dutch office account describes a weight-sensing smart fridge for a similar broad product range. Both public records explain a door-opening shopping flow, while identifying the detection systems differently.
Singapore: pre-authorise, open, take and close
The Singapore account places card or mobile-payment pre-authorisation before the door unlocks. The customer then takes products and closes the door. The source says the AI system identifies the selected goods and calculates the final amount.
The project is described as a multi-machine installation, but the exact unit count and mall name are not given. The public record does not disclose the installed camera arrangement, recognition model or approved product catalogue.
A recognition-accuracy percentage appears in the supplier's feature description. No test protocol, sample size or customer-site results accompany it. It is therefore not used here as a verified performance figure for this deployment.
The Dutch account names weight sensing
The Dutch office record describes card or mobile-wallet access, direct product collection and detection through changes in weight. It says products placed back before closing are excluded from the purchase.
This is a different stated detection method from the Singapore AI account. The records do not establish that the two installations use the same sensors, software version or shelf configuration.
The Dutch page discusses a payment hold and settlement, but its sequence is worded differently from the Singapore page. These accounts should not be combined into one universal banking flow. The payment processor, authorisation amount and release timing would need to be specified for an actual order.
Remote visibility appears in both records
The Singapore page lists stock, pricing, machine status and sales visibility through cloud management. The Dutch page lists stock tracking, low-stock alerts and sales by product. Neither publishes a detailed replenishment log or a support history from the customer.
These are reported snack-and-drink projects in a mall and an office. Their useful common detail is direct access to the products after payment access is accepted. Their useful distinction is the named recognition method.
For a buyer comparing AI and weight sensing, the next concrete inputs are the intended products and packages, shelf plan and payment market. The public records do not provide a controlled comparison between the systems, and their revenue forecasts are not customer operating results.
Project sources
