Where the days disappear
A request arrives as an email or PDF, often as free text, with item descriptions that almost match and quantities in the wrong unit. Someone on the inside sales team translates that into article numbers, finds the customer-specific prices and price tiers, and builds the quote. Complex requests need a senior salesperson, and they are already busy. So a quote takes days to turn round, while the customer has asked three suppliers at once.
What AI actually does here
The AI reads the request, recognises the products and matches them against your own catalogue, including alternatives when an item is out of stock. Prices, price tiers and customer terms come from the ERP rather than a stray spreadsheet. The result is a draft quote that the salesperson checks and sends. The knowledge held by your best salespeople (what suits what, which alternatives are common) gets captured along the way, instead of walking out the door when they retire.
Master data first, and that is not a detour
An honest word: AI on dirty article and price data produces dirty quotes. Duplicate article numbers, out-of-date prices and empty product fields turn up in every wholesaler. So the first and most rewarding step is sometimes to structure the catalogue. That sounds like a detour, but it is the same road: every improvement in your master data then pays off in every quote, every order and every stock decision.
Start narrow, prove it, expand
The approach that works: start with one product group or one customer segment, take the automation live there and measure the turnaround time. Usually a working first result is in place within a quarter. Then expand on the strength of what it proves, rather than on a grand plan drawn up in advance.