Set the cover
Choose the meals and portions this shopping cycle actually needs to provide.
All case studies
Client product delivery
From a family’s mental load to one clear household system.
The Family Fridge connects meal cover, recipe choice, portions, family input, food needs and the physical shop—without asking a household to become a project manager.
Production-test V1
Private household testing and final handover are continuing.
202governed recipes
65automated tests
3isolated household test environments
1connected workflow
The product problem
A shopping list is the output. The difficult part is everything upstream.
Household food planning is usually spread across memory, messages, saved recipes, dietary knowledge, leftovers and a list that changes aisle by aisle. Treating those as separate features simply moves the coordination burden onto the user.
The product was therefore designed as a single stateful loop. Each decision carries forward: required portions shape recipe selection; selected recipes shape quantities; household knowledge shapes the list; and the locked shop remains useful when the food is cooked.
The connected workflow
Choose the food
Search or filter the recipe authority, with household preferences and food needs in view.
Shape the list
Combine recipe quantities, recurring products, items already at home and manual additions.
Use it in the shop
Lock a stable shopping cycle, tick the trolley and keep the meals available afterwards.
The product in use
Four moments. One continuous decision trail.
Captured from an isolated demonstration household using fictional names and no client data.
Plan
01Quantity starts with real life
The household describes the food it needs before choosing recipes. Days guide quantity without pretending the meal plan is a rigid calendar.
Choose
02A recipe authority, not a loose bookmark list
A governed catalogue of 202 recipes supports search, structured filters, preparation details, portions, preferences and food-needs checks.
Build
03The recipes do the maths
Ingredients consolidate into one editable list. The primary user can account for food already at home, adjust quantities and add household products before locking the shop.
Shop & cook
04The plan survives contact with the supermarket
Shopping Mode freezes the agreed list, tracks trolley progress and keeps the selected meals ready for cooking. A controlled swap flow handles in-shop changes without silently rewriting history.
Product decisions
Built around the household, not around a feature checklist.
Two deliberately different surfaces
The primary user controls planning and sensitive household settings. Family members receive a simpler route for suggestions and meal reactions.
Food needs stay visible
Recorded allergies, intolerances and preferences inform recipe suitability. The system warns and asks for an explicit decision; it does not invent substitutions.
A stable in-shop state
Once shopping begins, the list becomes a durable cycle. Trolley progress, cooked meals and controlled swaps are recorded against that exact shop.
Authority before convenience
Recipes, units, pack behaviour, household variants and shopping provenance are structured so the interface can explain where quantities came from.
Delivery
Product thinking carried through into working software.
Constellation Highway handled the workflow design, interface, governed data model, shopping calculations, implementation, automated testing and deployment architecture as one delivery problem.
Privacy-conscious accessSeparated primary and family pathways
Stateful cloud architectureReact interface, Worker service and durable relational data
Evidence-led releaseAutomated tests plus isolated household verification environments
Current boundary
This case study shows the production-test product through a fictional demonstration household. Public access to the working application is being held back while private household testing and final handover continue.
Start a conversation
Have a problem, process or product that needs its next stage?
Bring the complicated version. We’ll define what needs to work, agree a sensible scope and identify whether the right next step is discovery, prototype, build or deployment.