Fourteen stores, one stock picture, and reorders the system suggests.

What was going wrong

A customer asks for a medicine at the Kothrud store. It is out. The Baner store, three kilometres away, has forty strips that will expire in two months. Nobody knows this, so the customer goes to a competitor and the forty strips go in the bin. Multiply by fourteen stores and a few thousand products.

Ordering is done store by store, from memory and from the supplier’s rep. Some stores over-order what sells slowly; others run out of what sells fast. The owner finds out about expiry losses when the auditor does.

What we would build

One live picture of stock across all fourteen stores and the godown, on the owner’s phone and on each store’s counter. When a store is short, the counter shows which store nearby has it, and a transfer is two taps: the sending store gets a pick list, the receiving store gets a confirmation, and nobody phones anyone.

Every morning each store gets a suggested reorder list: what is likely to run out this week, what is close to expiry and should move, what has not sold in ninety days and should not be reordered. The system suggests; the store manager approves, edits or ignores, and every decision teaches the suggestion for next week. Expiry alerts fire ninety days out, not at stock-take.

What we would agree up front

• Stock-outs. Times a customer asked for something that a store within five kilometres had on the shelf. Counted at the counter.
• Expiry write-offs. Stock thrown away, counted monthly, with the previous year as the baseline.
• Daily close. Minutes from the last bill to books closed, per store, so the owner sees the whole chain's day before dinner.

What the owner would be left with

One system in the chain’s own accounts, with each store’s existing billing kept as it is where that is simpler. Store managers trained one by one, in their own store. A suggestion engine they can override, and that gets better because they do. A written note at the end: which stores adopted it fastest, what the fifteenth store will need, and where we would look next.

This is AI integration on top of numbers you can trust: the reorder list is only as good as the stock picture under it, which is why the running costs are worked out before the build here too.