AStayed chilledArrival 4°C
12 hours at 4°C
- Quality
- Better preserved
- Remaining shelf life
- Potentially longer
FOOD SYSTEMS · ARTIFICIAL INTELLIGENCE · CURIOSITY
Good food.
Better journeys.
A little more intelligence.
What if we could understand a food’s journey
before it becomes waste?
I’m Zhong (Johnny). I explore how AI and better cold-chain information can help us ask better questions — and make more thoughtful decisions about food.
Try a little experiment No technical background needed.A FOOD’S JOURNEYSo much happens in between
+
Freshness has a history.
Let’s pay attention to it.
In 2022, 1.05 billion tonnes of food waste were generated worldwide.
01 / THE EXPLORATION TABLE
Small experiments.
Something to think about.
Move a slider. Change a decision.
See what a single number can hide.
COLD-CHAIN HISTORYTwo batches. Both 4°C.
Two batches. Both 4°C.
Not the same story.
Start with two batches of the same fictional food, in the same initial condition. Batch A stays chilled. Give Batch B a warm interruption, then cool it back to 4°C.
No interruption6 hours
Recorded temperature12-hour journey
Batch A · steady Batch B · interrupted
Both arrival readings4°C
Extra exposure above 4°C24°C·h
BCooled back downArrival 4°C
2 hours at 16°C, then back to 4°C
- Quality
- More degradation possible
- Remaining shelf life
- Potentially shorter
Cooling back down does not reverse degradation.
Same final temperature does not guarantee the same quality or remaining shelf life.
Conceptual comparison, not a shelf-life prediction. Actual differences depend on the food and its initial condition. Illustration colours are teaching cues; quality loss may be invisible.
MONITORING & EVIDENCEA quiet log.
A quiet log.
A missed chance to act?
A fictional shipment warms up between minute 35 and 55. Choose how often the sensor records the temperature. Longer intervals between measurements can miss the event — and the opportunity to investigate before more quality is lost.
5 minutes60 minutes
Temperature & sensor measurements2-hour window
Simulated temperature Recorded temperature
Simulated peak12°C
Highest recorded temperature4°C
No spike recorded
All samples read 4°C. The temperature still reached 12°C.
A chance to act is missed
Quality loss or shorter shelf life could go unnoticed.
Flag an evidence gap
AI could flag sparse coverage and ask for door events or another sensor.
Better evidence → more useful AI assistance.
Monitoring captures the signal. AI can help interpret patterns, combine context and prioritise review.
Illustrative workflow, not a running AI model. The response shown is determined by the simulated samples. Detecting a spike supports investigation; it does not establish product quality or safety.
THE SCALE OF PREVENTIONA small percentage.
A small percentage.
A lot of food.
Imagine a business handling 10,000 kg of food each week. What would a modest reduction in loss mean over a year?
0%10%
Small improvements deserve attention when they repeat across shipments, weeks and businesses.
Annual food kept in useIllustrative scenario
10,400kg per year
10,000 kg/week × 2% × 52 weeks
What these experiments show — and their limits
These are deterministic educational illustrations, not AI models, measured trials or food-safety tools. The first experiment sums temperature above an arbitrary 4°C baseline over time; it does not estimate freshness, shelf life or safety. The second uses a fixed triangular spike and samples starting at minute zero, so results depend on timing. The third is arithmetic: the selected percentage is an absolute share of throughput, not an observed or promised reduction. Its crates illustrate magnitude, not individual shipments.
Appropriate conditions depend on the food and process. The UNEP–FAO sustainable cold chains report explains the broader opportunity. These experiments illustrate why context and evidence matter before applying AI.
02 / THE PERSON BEHIND THIS
Hi, I’m Zhong (Johnny).
PhD researcher in AustraliaI’m curious about how AI can help us understand food systems — and how we can make that knowledge useful beyond a technical audience.
My interests include knowledge graphs and ontologies: ways to connect cold-chain observations and give them context. I’m also interested in education, emerging technology and making AI easier to understand. This is where I share things I’m exploring, building and learning.
Working with food, cold chains or AI? I’d love to hear what questions you’re trying to answer.