AI and automation for retail chains in India

The problem

Every store runs its own POS and stock list. The head office learns about stock-outs from customers and about excess stock from the annual count.

What we build

A nightly feed from every POS into one central stock table, with a dashboard for the purchase team and WhatsApp alerts when fast movers drop below their minimum in any store.

Expected outcome

Typically, a chain moves from a weekly manual stock count to a stock view that is less than 24 hours old, and purchase decisions stop depending on phone calls between stores.

How it works

1. Discovery: we map every store's POS (GoFrugal, Marg, Vyapar or a custom POS), the warehouse system and how stock moves between them today.
2. The store manager closes the day on the POS; n8n pulls each store's stock and sales export every night through the POS API, a scheduled export or a shared folder.
3. A central stock table is built in Google Sheets or a small database: SKU by store by quantity by last-sale date.
4. The purchase team gets a dashboard showing below-reorder items per store and stock that is idle in other stores.
5. The area manager receives a WhatsApp alert when a fast-moving SKU falls below its minimum in any store.
6. Barcode scanners at goods receipt update the same table, so the view stays within a day of reality.

Systems: POS software · Tally Prime · Google Sheets · n8n · barcode scanners · WhatsApp Business API

The problem

Store closing figures arrive as photos and voice notes. The head office retypes them, and the owner sees yesterday's numbers after lunch, if at all.

What we build

n8n collects each store's day-close from the POS, compares it to targets and last week, and sends a one-page report on WhatsApp to the owner and area managers at a fixed time.

Expected outcome

Typically, the head office stops retyping closing figures and the owner reads the day's numbers the same evening, with problem stores flagged.

How it works

1. The owner sets monthly targets per store in a Google Sheet; the accounts head lists the exceptions that matter (cash variance, high discounts, returns).
2. After day-close, n8n reads sales by store, by category and by payment mode (cash, UPI, card) from each POS.
3. The report compares each store with the same day last week and with the month-to-date target.
4. A one-page table goes to the owner and area managers on WhatsApp at 9.30 pm; each store manager gets only their own store.
5. An exceptions list goes to the accounts team: cash not deposited, stores that did not close, discounts above the limit.
6. Every Monday a weekly summary ranks stores and shows the top and bottom categories.

Systems: POS software · n8n · Google Sheets · WhatsApp Business API · Tally Prime

The problem

The buyer reorders from memory and from complaints. Fast movers run out before festivals, and slow movers are bought again because nobody checked last month's sales.

What we build

A weekly reorder list per supplier and per store built from sales velocity, lead time and minimum display stock, approved by the buyer and sent as a purchase order to the supplier and to Tally Prime.

Expected outcome

Typically, the buyer spends an hour a week approving instead of a day building lists, and stock-outs of top sellers become the exception rather than the festival-season rule.

How it works

1. The buyer sets minimum display stock and the lead time for each supplier; the system reads 8–12 weeks of sales per SKU per store from the POS.
2. Each week the system drafts a purchase list per supplier, split by store or by warehouse.
3. The buyer reviews the list on a simple screen and adjusts for Onam, Vishu, Diwali and planned promotions.
4. Approved purchase orders go to the supplier as a PDF by email or WhatsApp; a copy posts to Tally Prime as a purchase order.
5. On delivery, the store team scans barcodes; short supplies and wrong items are flagged to the buyer before the bill is booked.
6. A monthly report shows each supplier's fill rate and late deliveries so the buyer can renegotiate.

Systems: POS software · Tally Prime · Google Sheets · n8n · WhatsApp Business API · barcode scanners

The problem

A store runs out of an item that another store has in excess. The manager calls three stores, someone sends it in an auto, and no transfer note is made, so both POS stock counts are wrong.

What we build

Store managers ask the chain's WhatsApp number for an item. A bot checks the central stock table, suggests the nearest store, and creates a transfer note that both stores scan.

Expected outcome

Typically, transfers get a paper trail from day one and the stock count in each POS stays trustworthy, which is what makes every other report believable.

How it works

1. The store manager sends a message like “need 20 of item 4471” to the chain's WhatsApp Business number.
2. The bot checks the central stock table and replies with which nearby stores hold the item and how many.
3. The manager picks a store; a transfer note is created and the sending store's team gets a pick request.
4. The sending store scans items out and the receiving store scans them in; both POS stock levels update.
5. The area manager sees transfers that have been pending for more than 48 hours.
6. A monthly transfer report by store shows which items are being allocated to the wrong stores at purchase.

Systems: WhatsApp Business API · POS software · Google Sheets · n8n · barcode scanners

The problem

Apparel past the season, electronics past the model year and near-expiry pharmacy or food batches are discovered at the stock-take, when a markdown no longer helps.

What we build

A weekly ageing report per store and category, with markdown or transfer proposals that the owner approves in one tap on WhatsApp and that update the POS price automatically.

Expected outcome

Typically, ageing stock is acted on weekly instead of quarterly, and the owner can see the rupee value of old stock per store at any time.

How it works

1. The category manager sets age limits per category: for example 90 days for apparel, 60 days for electronics, 30 days before expiry for food and pharmacy.
2. Each week, n8n reads stock age per SKU per store from the POS and builds an ageing list with value at cost.
3. The category manager proposes a markdown or a transfer to a store where the item still sells.
4. The owner approves on WhatsApp with one tap; approved prices are pushed to the POS and a promo tag is printed at the store.
5. Two weeks later the system reports whether the marked-down items sold.
6. For pharmacy and food, near-expiry batches go to the store manager 60 and 30 days ahead for return to the supplier.

Systems: POS software · Google Sheets · n8n · WhatsApp Business API · Tally Prime

The problem

Sales are entered into Tally Prime by hand at month end. HSN codes and GST rates differ between POS and Tally, UPI settlements do not match the bank, and the CA reconciles for days.

What we build

A nightly job posts each store's sales and payment-mode totals to Tally Prime, matches UPI and card settlements to the bank statement, and flags GST mismatches before the return.

Expected outcome

Typically, month-end posting disappears and the CA gets a clean, reconciled sales ledger within a day of month close.

How it works

1. The accounts head and the CA map POS items to Tally stock items with HSN codes and GST rates; mismatches are fixed once.
2. Each night, n8n reads POS sales and payment-mode totals and posts sales vouchers to Tally Prime, per store per day or per invoice as the CA prefers.
3. UPI and card settlements are matched against bank statement entries; variances are listed for the accounts team.
4. B2B invoices (bulk buyers, corporate orders) are flagged for e-invoicing if the chain is above the ₹5 crore turnover threshold.
5. Supplier purchase bills are scanned with OCR and matched to purchase orders before posting.
6. The CA receives a GSTR-1-ready export and an exceptions list each month.

Systems: POS software · Tally Prime · n8n · bank statement exports · OCR

The problem

Offers and order-ready messages go out from a staff member's personal number. The number gets blocked, replies are missed, and there is no opt-in or opt-out record.

What we build

Official WhatsApp Business API with approved templates, opt-in captured at billing, triggers from POS events, and replies routed to the right store.

Expected outcome

Typically, the chain moves from an unblockable but untracked blast to a messaging channel it can measure, with opt-in records that stand up to scrutiny.

How it works

1. Opt-in is captured at billing through a POS prompt or a QR code at the counter; the number and consent date are stored.
2. Templates are approved on the WhatsApp Business API: order ready, alteration done, points balance, store-specific offer.
3. Messages are triggered by POS events (bill, alteration, pre-order) or by a campaign list filtered by store and last purchase date.
4. Replies go to the store's shared inbox; a “STOP” reply removes the customer automatically.
5. A dashboard shows sent, delivered, read, replied and redeemed per store and per campaign.
6. Meta's per-message charges are tracked per campaign so the marketing head can see the cost per redemption.

Systems: WhatsApp Business API · POS software · Google Sheets · n8n