Larder
Product Design
Problem: small kitchens count stock on paper and order on WhatsApp, where a “seen” tick is not a confirmation.
What I did: desk research, a head chef's journey map, a desktop and mobile concept, and a heuristic review of my own screens.
Outcome: one loop from count to delivery. 10 issues found, 3 major, all fixed. Not yet tested with kitchen staff.
Larder is a concept for a stock and ordering tool for small independent restaurants in Portugal, with one to three venues. Chefs see what is below par, send supplier orders on WhatsApp the way they already do, and check what went to waste each week.
Role:
Product Designer
Year:
2026
Explore the full story –
Challenge
Restaurants are Portugal's second-largest source of food waste after households, with more than 237,000 tonnes in 2020 (INE). In small kitchens, stock lives in a notebook and orders go out on WhatsApp, where a “seen” tick is not a confirmation. Full inventory tools start at around USD 249 a month and assume a chain's setup time.
Objective
A tool a small kitchen can start using in minutes: a phone count by storage area, an order suggested from par levels and sent on WhatsApp, a confirmation step that closes the loop, and a weekly waste view tied back to the par.
Results
Desk research and the journey map shaped one loop: count, order, confirm, check in, review waste. A heuristic review of my own screens found 10 issues, 3 of them major, all fixed and shown before and after. Not yet tested with kitchen staff.
Key insight
In a small kitchen the order itself is easy; what fails is everything around it: counts of different ages, a WhatsApp tick that is not a confirmation, and deliveries checked from memory.
Process
Discover. INE and Eurostat food waste data, and an audit of seven inventory tools and their public reviews.
Map. A head chef's Friday, from count to waste, marking where things slip and where Larder steps in.
Design. The user flow first, with a way back at every decision, then desktop screens for the office and mobile screens for the kitchen floor.
Review and fix. A heuristic review against Nielsen's 10 heuristics found 10 issues. The 3 major ones: below-par numbers built on counts of different ages, a count that could finish with items left, and sent orders with no deadline for a reply.
Success metrics
These are the targets I would use to judge a pilot. They are goals, not measured results: the product has not been built or tested.
Weekly stock count done in under 15 minutes on one phone.
At least 90% of supplier orders confirmed within 24 hours.
Food waste cost down 15% after three months, compared with the first four weeks of logging.
No item runs out without first being flagged as below par.
Shipped & next
Left out on purpose: recipe costing, POS integration and transfers between venues. They matter to chains more than to one-to-three-venue kitchens.
Next: five 45-minute sessions with chefs of small Lisbon restaurants, on a real phone in their kitchen. Tasks: finish a count with one area left, send Friday's order, mark a supplier reply, check in a delivery with one item short. Success: every task done without help.