Introduction

(by Adam Wilkinson)

We wanted to write a piece about the latest version of Meal Analyser referencing how we used the project as an internal learning exercise in integrating AI. It has exceeded all our expectations and convinced us that it is possible to use AI for public good in a practical way. It also taught us that the current AI is capricious, never says “I can’t do it”, just guesses but does not tell you; and will suddenly decide to do a simple piece of maths wrong. But the power is extraordinary, and the importance of the use of words, grammar and language a fascinating change from many years of maths, logic, and code. Anyway, we wrote the article and then I could not resist letting AI have a go with it. This is it, awful but impressive. The brief I gave it was to take my original and rewrite it in a form that would be good for a LinkedIn piece. We have not changed anything, even where we disagree with it. So if you want to “take it for a spin’’ then please do. If you want to read my original just contact me.

Using AI to Build the New Meal Analyser (by AI from Adam original)

Transforming food service carbon accounting with simplicity, scale, and rigour

We’ve just released the next generation of Meal Analyser—our AI-powered tool for calculating and optimising the carbon impact of meals.

Originally built during the Strength2Food Horizon research project, it was designed to help anyone—from school caterers to large food service providers—understand the climate impact of what they serve. The early version was free and easy to use, but to gain better accuracy, we developed a deeper dish-level analysis with the University of Edinburgh.

It worked—too well. The system became more expensive and time-consuming, moving beyond the reach of many users.

Back to First Principles

Six months ago, we revisited our core goals:

✅Simplicity
✅Accessibility
✅Academic rigour

The biggest challenge? Manual data gathering and coding—time-consuming and often incomplete. That’s where AI changed everything.

A New Model for Food Carbon Impact

Meal Analyser now works through a 3-way partnership between:

🔹 The organisation – providing basic meal inputs
🔹 AI – sourcing ingredients, matching them to emissions (Agribalyse database), and adding nutrition, and calorie analysis
🔹 Our team – ensuring that the maths is consistent, traceable, and science-led

It seamlessly scales vertically (dish → service → organisation) and laterally (adding nutrition, calorific values, and disposal emissions). And it’s still free to try.

What It Looks Like

Here’s a real example: Chicken and Cheese Pasta Bake

This dish emits 4.29 kg CO₂e per kg—medium intensity. But with AI-driven optimisation (like switching proteins or cutting waste), we can bring that down significantly.

What We’ve Learned

🚀 AI gives a massive productivity boost
🧠 LLMs sharpen communication—we’ve learned to think in code and words
🧮 AI is excellent at context but needs solid logic frameworks and control for maths
🌍 At a small scale, we’ve shown it can work—for public good

Try the Demo – It’s Free

We’re offering free credits to test the new Meal Analyser. If you work in:

🍴 School Meals
🏥 Hospital Catering
🥗 Universities
🚛 Contract Catering
📊 Local Government

…or any area where food meets climate targets, we’d love you to take it for a spin.

🔗 Try the Demo Now

#FoodServices #ClimateAction #AIforGood #CarbonFootprint #SustainableFood #PublicSectorInnovation #MealAnalyser #ImpactMeasurement #GHGEmissions #Horizon2020

If you need more information or have any questions don’t hesitate to contact us directly.

Contact Us