About this case study: it is based on a business typical of the ones we work with, not a named client, and every number comes from the assumptions at the end of the page. Swap in your own and the sums still work.
The problem
The business sells sarees and dress material to around 300 dealers in Maharashtra, Madhya Pradesh and Rajasthan. Almost every order arrives on WhatsApp: a typed list, a photo of a page from the dealer's notebook, or a voice note. Design numbers are easy to misread, and a wrong design means a return a week later.
Three people on the order desk retype each order into a Google Sheet, then into Tally as a sales order. Once goods leave, dealers call to ask for the LR number and the transporter's name, and someone looks it up and calls back.
What we build
- 1
Orders read for you
Orders keep coming to the same WhatsApp number. An AI step reads text, photos and voice notes and turns each one into a draft order with design number, quantity and rate, checked against the item list.
- 2
One-tap approval
The desk sees the draft next to the original message and approves or corrects it. Approved orders go into the sheet and into Tally as a sales order, with nothing retyped.
- 3
Dispatch updates without calls
When the dispatch team enters the LR number, the dealer gets a WhatsApp message with the LR number, transporter and number of bundles.
- 4
Morning pending list
At 9:30 every morning the owner gets a short message with pending orders by dealer and anything waiting more than two days.
Before and after
| Step | Before | After |
|---|---|---|
| Logging one order | About 12 minutes | About 2 minutes to check |
| Dispatch status calls | About 20 a day | A few, for unusual cases |
| Pending order list | 40 minutes, by hand | Sent automatically |
Where a person stays in control
Nothing goes into Tally until a person approves the draft. Orders the AI is unsure about (an unclear photo, an unknown design number) are flagged instead of guessed.
Running cost
Roughly ₹6,000–9,000 a month for WhatsApp Business API messages, AI usage and hosting, depending on order volume.
Tools
- WhatsApp Business API
- Google Sheets
- Tally Prime
- n8n
- AI model for reading photos and voice notes
How the numbers work
- 35 orders a day, 26 working days a month.
- Each order takes 12 minutes by hand and 2 minutes to check: 35 × 10 min × 26 days ≈ 152 hours.
- 20 dispatch calls a day at 3 minutes each: 20 × 3 min × 26 days = 26 hours.
- Staff time costed at ₹350 an hour (salary plus overheads): 178 × ₹350 = ₹62,300.
Estimates, not promises. Use the savings calculator with your own hours and costs.
Want this for your business?
A free 30-minute call is enough to tell whether it fits, what it would take and what it would cost.