AI grocery intelligence for households that throw away £800 of food a year.
Built during MSc Management at Bayes Business School and qualified for VC funding at a Bayes entrepreneurship event. The wedge is zero manual entry — receipt AI reads the shop, computer vision tracks expiry, a recipe engine spends whatever is about to spoil.
- Receipt AI
- Expiry computer vision
- Recipe engine
- Zero manual entry
UK households lose roughly £800 a year to food waste, 9.5 million tonnes nationally. The cause is invisibility rather than carelessness: no real-time view of what is in the house, expiry dates nobody is tracking, duplicate buying, and dinner decisions made without reference to what is already in the fridge.
Existing trackers already solve this on paper — and are abandoned within a week, because every one of them requires manual entry. The failure is not the tracking model, it is the input cost.
Ran primary research before building anything, March 2026. 75% cited over-buying as the top waste driver, 71% would actively use the app, 64% have food expire unnoticed, and 57% said lack of time makes it worse. 82% of respondents were 18–24.
Designed the flow as three steps with zero manual entry: AI reads the shopping receipt and logs every item, computer vision reads expiry dates off packaging and schedules reminders, and the recipe engine works from live inventory, prioritising whatever expires first.
Freemium — a free tier capped at three scans a month, premium at £3–5. The price came from the research rather than a guess: 63% were willing to pay £3–5 a month and 35% indicated £5+.
Beachhead on university campuses for the lowest CAC and the highest word-of-mouth velocity — one flat is four users. Roughly £180M serviceable obtainable market inside a £4.5B serviceable market of tech-adopting households aged 18–45.
Three phases: validate an iOS and Android MVP in closed campus beta (months 0–6), freemium App Store launch with an ambassador programme and premium go-live (6–18), then supermarket API integrations, predictive replenishment and a carbon-footprint layer (18–36).
Software-only delivery of smart-fridge-level intelligence at zero hardware cost — the one position manual trackers, recipe apps and smart fridges each cannot hold at once.