Where they started
A wholesale meat distributor. Customers ordered the way that suited them, by text message, WhatsApp and several other channels. Every order then went through the same loop. Someone read the message, worked out which products were meant, looked each one up on the price list and keyed the sales order. The quote and the invoice followed from the same typing.
Customers use their own names for products and prices change. The lookup took longer than it sounds. The number of orders the business could take in a day was set by how fast people typed.
What we built
An agent connected to their product database. It reads the text and WhatsApp messages as they arrive, works out which products and quantities the customer means, prices each line from the database and drafts the sales order and the quote. The price list check that used to be a person with two screens open is now the first thing the agent does.
Nothing changed for the customer. They still send a text or a WhatsApp in their own words.
There is one price list and it is the one the business already maintains. The agent prices every line from it. Messages are messy because customers write the way they talk, and the agent reads them.
Where it landed
The time spent reading messages and keying orders went away. The business now handles more than 100 orders a day, and it has doubled without adding headcount. That is the ordering and quoting problem most wholesalers have, with the extra wrinkle that the orders arrive as chat messages.
Why we started with the database
The hard part of most order work is the data. Here the product database already existed and was kept current. The agent had something reliable to price from on day one. That is where we always look first in a workflow automation build. If the price list had lived in three spreadsheets, the first job would have been fixing that.