Fresh produce AI case study: wholesaler speeds up orders and cuts berry waste
A berry and leafy greens wholesaler cut order entry from about 8 hours a day to 2, and reduced dumped stock from 6% to 3.5% of purchases, after adding Pak-Bot.
At a glance
| Measure | Before Pak-Bot | After |
|---|---|---|
| Order entry time (hrs/day) | 8 | 2 |
| Stock dumped (% of purchases) | 6.0% | 3.5% |
| Customer credits (% of sales) | 1.8% | 0.9% |
| Time to trace a failed QC lot to customers | About 2 hours | Under 1 minute |
The client
The client wholesales berries, leafy greens and herbs from one warehouse with about 45 staff. It buys from around 30 growers and delivers to roughly 300 restaurants, independent grocers and food service customers.
The challenge
- Orders arrived by email and were retyped. About 120 orders a day came into a shared Gmail inbox; typos caused wrong or short deliveries.
- Stock went off in the cool room. Older lots were often picked after newer ones; about 6% of purchased stock was dumped.
- Credits ate into margin. Wrong products, short shipments and quality complaints cost about 1.8% of sales in credits.
The solution
Pak-Bot inside Producepak works on live data with no exports or sync delay, and existing site permissions stay in place.
Orders straight from email
Pak-Bot checks the Gmail inbox and creates orders; the office reviews and approves instead of typing.
Accurate promises
"Can we fill an extra 200 punnets of strawberries tomorrow?" is checked against stock and incoming deliveries.
FIFO at the point of picking
A newer lot scanned while an older one is in the cool room is flagged before it goes on the truck.
Expiry warnings in time to act
Sales is warned which lots are close to expiry, so they are discounted instead of dumped.
Quality linked to deliveries
"Which customers got the lot that failed QC this morning?" returns the list in seconds, with photos.
Results
| Benefit | How it was estimated | Annual value |
|---|---|---|
| Less stock dumped | Waste fell from 6.0% to 3.5% of about AU$6M purchases | AU$150,000 |
| Fewer customer credits | Credits fell from 1.8% to 0.9% of about AU$7.5M sales | AU$67,500 |
| Order entry time saved | 6 hrs/day × 260 days × AU$40/hr | AU$62,400 |
| Total | AU$279,900 (illustrative) |
One of the two order entry staff moved into customer service, following up late payers and new accounts. Customers noticed fewer wrong deliveries and faster answers to complaints.
Why it worked
It removed retyping
Orders go from the customer's email into Producepak without being keyed again.
It acts on live stock
Answers reflect the cool room as it is right now.
It joins quality and logistics
A failed inspection leads straight to the customers and pallets affected.
Put a question to Pak-Bot about your own packhouse
Book a session with a Producepak consultant. You will see Pak-Bot working on real produce data, and talk through your crops, sites, customers and documents.