SmartCart, a UX and AI case study

What’s coming up for you?

SmartCart is a concept for planning an occasion from the supermarket. This is the story of how it came about: the problem, the idea, the moment I learned it wasn’t new, and the decisions that followed, with the reason each one is better than the alternative I considered.


Project

Self-initiated concept

My role

UX and AI product design

Platform

iOS and Android

Scope

Research, concept, flows, UI

Status

Not yet tested with users

The Short version

I started with a brief about supermarket frustration and ended with a question about trust: when an AI builds your shopping list, how do you check it, and what happens when it’s wrong?

SmartCart is a concept app that answers it. You say what’s coming up: a birthday party, a trip, a school project. An AI drafts the whole list across food and non-food, filtered for dietary needs. You review it, accept swaps for anything out of stock, then shop it in the store along an aisle route or send it for delivery.

Where this stands. This is a concept. The statistics come from published studies, linked in the sources. The personas are illustrative, built from those studies rather than from interviews I ran. The screens haven’t been tested with users yet; “What I’d do next” says how I would.


The route I took

The order mattered, so here it is, with the reason each step sat where it did.

  1. Framed the brief as a sentence : I wrote the problem the way a shopper would say it. Everything after depends on that sentence, so it came first.

  2. Read the published research : I had no interviews, so I used what others had measured, and checked every statistic against its source.

  3. Built five personas : Composites from that research, spread across countries, so no single kind of shopper drives the design.

  4. Changed the question : From “what do you want to buy?” to “what’s coming up for you?”, which led to planning at home and treating the store as the fallback.

  5. Split the work between AI, rules and the shopper :The AI drafts, fixed rules handle anything that must not be wrong, and the shopper confirms. I decided this before drawing a screen, so safety never depends on a model’s guess.

  6. Mapped the journey and sketched wireframes : A flow on paper is cheap to change, so I tested the order of steps before any visual design.

  7. Checked the market, twice : Once while the concept formed and again before publishing. The second check changed the positioning.

  8. Designed in Figma and reviewed line by line : Each round fixed small things, like a button that didn’t say what it did or a count that didn’t match.

  9. Planned the tests : Hypotheses to try, and a list of what is still unfinished.

It started with a queue

The brief was ordinary: make supermarket shopping less frustrating for people around the world. So I read what had been published about where it goes wrong, and checked every figure against its source.

Queues, empty shelves and moved aisles

In a 2018 survey of 4,000 UK shoppers by the consultancy Newton, 48% named slow checkouts as their biggest gripe, 46% named out-of-stock products and 38% named difficulty finding items. 65% were frustrated when a store changes its layout. More than a quarter (27%) said they wouldn’t buy an alternative in the same store if their preferred brand was missing.

Most food waste happens at home

The UN Environment Programme’s Food Waste Index Report 2024 found that 1.05 billion tonnes of food were wasted worldwide in 2022. Households accounted for up to 60% of it, about 79 kg per person each year. High-income and lower-middle-income countries differ by only about 7 kg per person, so this isn’t a rich-country problem.

Allergen labels are hard to use in the moment

A 2023 meta-analysis of 105 studies (3.3 million participants) put food allergy prevalence at 4.3% overall, with 4.8% in Europe, 4.2% in Asia and 7.5% in Oceania. In a GS1 UK survey of visually impaired adults, 41% had bought food and found later that it contained an ingredient they avoid, and 37% had had an allergic reaction because they couldn’t read a label. That survey shows what happens when label reading isn’t possible; it isn’t a picture of all shoppers.

Shopping gets harder as independence drops

The same GS1 UK survey (December 2025, 500 visually impaired UK adults) found that 71% of those with severe sight loss find food shopping a challenge, and 83% said packaging information is hard or impossible to access. In a Japanese survey of 2,028 older people living alone, 14.6% of men and 21.7% of women found shopping difficult; poor eyesight and having nobody to help were among the associated factors.

The findings point the same way: most of the trouble is decided before the shopper reaches the shelf.

What they planned, what they knew was in stock, and what they could safely eat. That was the thread I followed.

These studies come from different countries, years and sample sizes, and most are surveys. They show where friction is common. They don’t tell me how large the effect would be in any one store, and I haven’t run research of my own.


Previous
Previous

DETOX - Waste Management Application

Next
Next

Hilton Hotels & Resorts