Store and area managers at an Israeli retail chain were run off their feet — and blind to their own numbers, waiting days on HQ analysts for a spreadsheet. We put the data in their pocket, in plain language.
The gap between head office and the floor was costing the chain. Managers were working flat out, but when it came to data they were completely in the dark — and waiting on someone at HQ to hand it over.
Every time a store manager wanted to know which product was stuck in the back or why a department's sales had dipped, they opened a ticket and waited two or three days for a complicated spreadsheet from HQ — total dependence on the analysts and IT. The data existed, in the ERP, the tills, the time clocks, but a manager running between shifts, suppliers and customers has no time to dig through graphs. And issues — a fridge drifting out of temperature, a line missing from a shelf, a recurring customer complaint — were scattered across the call centre, surveys and messages, and only got handled once the damage had grown.
No new software to learn — just the tools they use every day, made to answer.
Rather than ask staff to learn new, complicated software, we connected an AI engine — wired only to the company's secure database — straight into WhatsApp and the staff app they already live in on their phones.
Instead of requesting a report, a store manager simply types: "which five dry-goods lines have sat in my back room over a month?" or "how many shifts am I short next week?" — and gets a precise, data-backed answer in seconds.
The assistant runs in the background over customer feedback and operations logs, spots trends on its own — a recurring complaint about a faulty product, an issue at a specific branch — and pushes the area manager an immediate alert with a suggested next step.
The engine is connected exclusively to the company's secured data, so answers come from the real records and nothing sensitive leaks — the freedom to ask anything, without opening the data up.
Because it lives inside WhatsApp, there was nothing to roll out and nothing to train — a manager who can send a message can query the whole operation. Usage was immediate.
With answers on tap, managers stopped queuing for HQ and went back to running the shop floor and solving problems for customers — decisions made on data, in the moment, not two days late.
The managers stopped working blind, and the analysts got their time back.
Time to get an answer and act on a problem on the floor fell from two or three days to a few minutes. Maintenance and operational issues were resolved about 25% faster, lifting both the shop floor and customer satisfaction.
Requests to the central data-and-analytics team dropped by around 70%, handing the organisation back the expensive hours its analysts had been spending building one-off spreadsheets for the field.
Two choices made it stick: it lives in a tool people already trust and use, so adoption needed no push; and it can only see the company's own secured data, so the managers got freedom to ask without anyone risking a leak. Built for the same client as our real-time pricing work, it's another piece of the same idea — putting production AI exactly where the work already happens.
Reef TRH