ScanSavr

The problem

Supermarkets can see everything up to the till, then nothing. Once food leaves the store, retailers can’t see how it’s stored, eaten or thrown away. Yet households produce around 70% of UK food waste, the industry loses roughly £1 billion a year to it, and the UK target is to halve food waste by 2030. Existing tools work before purchase, on logistics, or after the waste has happened, on redistribution. We aimed at the gap in between.

The idea

ScanSavr lives inside a supermarket’s existing loyalty app, starting with myWaitrose, rather than asking shoppers to download something new. Shoppers get expiry reminders, storage tips, and loyalty points for tracking what they use. Supermarkets get anonymised post-purchase data, to forecast better and waste less. Our primary research was with Waitrose’s Ethics and Sustainability Specialist, a relationship I brought to the team. She called direct app integration “the holy grail”.

My part: the technology

As Technology Director, I wrote the technology strategy and designed the app mock-up.

• No AI, on purpose. Tracking dates and timing reminders is structured, rule-based work. A deterministic rules engine does it with a fraction of the computing power of deep learning, so a retailer cutting its waste emissions (Scope 3) doesn’t add to its electricity emissions (Scope 2) along the way.
• No new hardware. An over-the-air update lets the scanners already in store read 2D barcodes added at packaging, capturing each product and its exact expiry date.
• Input, transformation, output. Data goes to AWS in London, the rules engine works out each item’s perishability, and REST APIs push a nudge into the app, such as a recipe idea 48 hours before something expires.
• Nudges, not restrictions. Drawing on nudge theory, the app offers helpful prompts at the right moment and rewards them with points. It never limits what anyone buys.
• Private by design. No personally identifiable information is processed: shoppers appear only as hashed tokens. Zero-trust architecture throughout, with TLS 1.3 in transit and AES-256 at rest.
• Ready for Christmas. Auto-scaling handles the peak trading weeks, and a microservices design is ready for a second retailer.

I also planned the team’s own secure setup: Google Workspace, Zoom, Cisco Duo for multi-factor sign-in, and Tailscale in place of a traditional VPN.

The plan

Waitrose as the first customer, with Tesco to follow within the first year. Pricing of £0.40 per active shopper, plus an £80,000 annual licence per supermarket. The plan sought £50,000 for a 14% stake and forecast break-even by July of the first year, with profit growing from £56,659 in Year 1 to £434,035 by Year 3. The other five members of the team led the market analysis, marketing, operations, HR and legal, and finance.

The result

The group report scored 80%, and my technology section scored 90%. The moderator called it “an outstanding and highly investable ScanSavr submission”, singling out the Waitrose research and the decision to reject energy-intensive AI. On my section, the marker’s verdict was simple: “This is exceptional work!”

The challenge they set was a fair one: whether the budget and team were big enough for such fast growth, with enterprise sales, software development and reliable support all needed at once.