Built in Manchester · For UK food brands

Everything your food business needs. Powered by AI.

One connected dashboard for menu intelligence, nutrition, health scoring, compliance, content and customer transparency - grounded in inspectable nutrition science.

Grilled chicken and avocado salad
Live food intelligence
Nutrition · health · marketing

7

Connected MVP modules

14

UK allergen categories surfaced where ingredients match

MSc

Nutrition-led, explainable design

Nutrition & Health Scoring

Edit the sample recipe or add recognised ingredients. Every score below is recalculated in this browser; no recipe is uploaded or retained.

Estimate, not a legal label. Values depend on the amounts and ingredients entered. The engine flags incomplete or high-risk cases for human nutrition review.

Your dish

Food Quality Score · per serving

Analyse a recipe to see results.

Nutrition profile

Live visualisation of the calculated serving. Chart values change with ingredient amounts and preparation.

Why this score?

A transparent weighted model, not a trained machine-learning model.

Analysis signals will appear here.

Compliance & dietary checks

These checks assist review; they do not replace recipe verification or legal labelling advice.

Allergen, dietary and review signals will appear here.

Health score signal mix

Positive and negative weighted contributions are visualised from the real scoring calculation.

Make the menu better

Analysis will identify input-led options here.

Verified-fact content studio

Content is template-based from the computed result. It makes no medical claims and is editable before use.

Analyse a recipe to prepare content.
Same engine, customer-facing output. This view uses the exact recipe analysis created in the business journey above.
Transparent food information

Analyse a dish first

Ingredient-led nutritional information, designed for clearer choice.

What’s in this serving

Dietary & allergen information

No analysis yet.

How we calculated this

The business supplied the ingredients, amounts and cooking method. Nutrient estimates are calculated from the embedded demonstration reference dataset and must be checked against the final recipe before publication.

Omnichannel marketing studio

Premium food campaign imagery

Create content from the same analysed dish. The copy is rules-based and nutrition-aware, so it will not invent taste, health or campaign claims.

No live posting is claimed or available in this MVP. A human reviews the output before publishing.

Marketing readiness

Channels ready
0
Fact coverage
Review status

All values derive from the current recipe input and campaign selections.

Campaign pack

Analyse a dish, then create a campaign pack.

Channel fit

Computed channel checks will appear here.

Marketing economics

Estimate time saved by preparing one fact-bound content pack instead of manually rewriting the same menu facts across channels.

No estimate calculated yet.

Customer value calculator

Use your own inputs. This model compares the plan’s £99/month subscription against estimated time and provider spend avoided; it is not a promise of savings.

Plan-derived economics

These are calculations from the business plan’s stated £99 average monthly price, 45 Year-1 customers and £8,640 cost of sales.

MetricComputed value
Year-1 subscription revenue
Gross margin on plan revenue
Break-even customers at £1,930 fixed cost and 84% margin

Estimated monthly value

Time value + avoided spend

Monthly net value

Estimated value less £99 subscription

Return on subscription

Net value ÷ £99

Learning loop: reviewer calibration

This is not ML training. It records a reviewer’s direction of travel only in this browser session, then adjusts the score’s calibration offset. It demonstrates the planned feedback-and-validation process without claiming a trained model.

No reviewer feedback recorded. The score uses a zero calibration offset.

Reviewed cases

0

Calibration offset

0

score points

Agreement rate

Based only on this session’s reviews

Data minimisation test

Paste text containing personal details and verify how the local redaction preview removes them before any analysis. This demo does not transmit or store the original text.

Nothing entered.
Observable limit: this is pattern-based redaction, not identity verification. It may miss information not matching the patterns, so operational controls and human review remain necessary.

Scalability measurement

Run a real local benchmark of the same scoring function using synthetic recipe variations. Result depends on this device; it is evidence of MVP compute throughput, not a production-scale claim.

No benchmark run yet.

Production safeguards to validate next

Verified nutrient database integration and recipe validation workflow
Role-based accounts, audit log and retention controls
Legal review of UK allergen and calorie-labelling outputs
Consent-led outcome data and model-validation study

These are deliberately not represented as complete in this offline demonstrator.