Case study
AI quality inspection across seven packhouses
A grower, packer and distributor of about 36 high-care, short-shelf-life crops replaced paper QC with Producepak QI. Inspection time per sample fell 70% and AI accuracy reached 99.8% within six months.
Free setup & training · Human support by named fresh produce consultants
At a glance
The challenge
- Paper QC forms and visual checks against printed specs
- 14.5 minutes per sample, causing queues at unloading bays
- Different results between inspectors and sites – inconsistent grower grading and occasional retail rejections
- Four full-time QA inspectors per site across shifts
- Reject reports retyped into the central system – supplier disputes delayed up to 24 hours
The solution
Producepak QI, a progressive web app with a custom computer-vision model, running on standard smartphones with no installation. The operator selects the crop, photographs the sample (for asparagus: both sides of the spear), the AI scores each criterion – blight, purple spot, rust, size conformity, insect damage, packaging integrity – and the inspector approves or overrides.
Want results like these in your business?
Cut produce waste – see how it worksRollout
| Month | Time per sample | AI accuracy | Status |
|---|---|---|---|
| 0 | 14.5 min | 81.0% | Pilot at one site |
| 1 | 8.7 min | 88.5% | 3 sites |
| 2 | 6.5 min | 93.5% | All 7 sites |
| 4 | 4.35 min | 98.2% | Labour savings achieved |
| 6 | 4.35 min | 99.8% | Near-zero corrections |
Business impact
- QA staff reallocated to packline optimisation
- Failed deliveries reported to purchasing and growers instantly, with annotated photos
- Shorter time in warm receiving bays protects shelf life of rocket, watercress and herbs
- QC data, Pak-Bot and Power BI used to predict quality and quantity before harvest arrives and to negotiate better supplier prices
- QC linked to the farming app and ERP inventory
In their words
"Our packhouse team takes photos of both sides of the asparagus stem, and the AI ticks off the criteria line by line in seconds."Quality Assurance Director
"By month two, the AI was demonstrably more accurate than our human inspectors. By month six, employee corrections dropped to near zero."Operations Manager
Related
- Fresh produce quality control software with AI inspectionFresh produce QC software with AI quality inspection on a smartphone: 70% faster inspections, 99.8% AI accuracy…
- How to reduce fresh produce wastePractical ways to reduce fresh produce waste: lot ageing, FIFO at the scan, expiry warnings, pack-to-order, QC at…
- Multi-site vegetable packer halves stock write-offsA three-site potato, onion and carrot packer cut reporting from 14 to 3 hours a week and stock write-offs from 3.5% to…
- Citrus and grape exporter cuts rejected containersA citrus and table grape grower-packer cut rejected export containers from 3 to 1 a season and mock recalls from 6…
- Produce wholesaler speeds up orders and cuts berry wasteA berry, leafy greens and herb wholesaler cut order entry from 8 to 2 hours a day, dumped stock from 6% to 3.5% and…
Sources and further reading
Questions
Is this a real Producepak customer?
Yes. Results are as published by Producepak; some customers asked for their names to be withheld.
How were the savings estimated?
From the before-and-after figures in the table, using the rates shown. They are the customer's estimates, not guarantees.