Bringing the taste of home to young Africans in Canada

Where it started

Arrive in Canada as an international student and one question follows you everywhere: what do I eat? For African students, it is rarely simple. Uber Eats and DoorDash serve everyone, which means the food they grew up on is sometimes hard to find.

The group was also growing fast. Between 2021 and 2023, study permit applications from Africa roughly tripled, from 90,000 to 273,000.¹ By fall 2023, close to 72,500 African students were enrolled at Canadian institutions.¹ Yet the apps they relied on for food were not built with them in mind.

I did not want to design on assumptions, so I started with a survey of 116 students who identify as African, across Canada, then ran 10 interviews to understand what the numbers could not explain.

My answer was a concept app that connects students to familiar food businesses, makes bulk ordering easy, brings recipes from home into one place and builds community around shared food. I took it as far as a high fidelity prototype and a brand system.

A prototype only tells you what people say they would do. Coming back to it, I wanted to see what they would actually use, so I set out to turn the concept into a working app real people could install.

The concept

The app today

Why young adults, not students

The first concept was for international students. I widened it to young adults of African origin, for two reasons. International student numbers fell about 29% between 2023/2024 and 2025/2026,² and the need is the same for anyone building a life far from home. Africa makes up 15.6% of recent immigrants to Canada.³ My testers bear it out: most have lived here 12 to 20 years and still miss the food.

Today the app has four parts.

• RecipesDishes from across the continent, step by step, with Cook Mode.
• GroceriesAfrican grocers and supermarkets nearby, nearest first.
• CateringCaterers for parties and events, with packages you can order in the app.
• Rootine PlanA meal plan built around the cuisines you love, from tested recipes first.

The first build

The first build went to five people across the Greater Toronto Area, in Mississauga, Markham and North York, through TestFlight. I kept the core and cut the commerce, to learn whether people wanted the app before building checkout.

Testing then showed that the thing I cut, ordering in the app, was what people missed most. The original research had hinted at it. That tension drives the rest of this story.

Testing

I put the first build in front of five people in one hour sessions, each following the same script: background questions, five live tasks while thinking aloud, then questions about features and price. After each task they rated how easy it was, from 1 to 5.

The five were different in the ways that matter. Names are changed.

• Funmi, from Lagos, 14 years in Canada. Cooks a fixed weekly plan but visits an African store only once or twice a year.
• Kola, a passionate home cook. Bulk shops every two weeks at Asian supermarkets because they are cheaper.
• Lina, from Jordan. Knows African food socially, shops at Loblaws and an Arabic store, and is trying meal prep delivery.
• Daniel, Nigerian Canadian. Rarely cooks and spends over $200 a month on Uber Eats, but wants to cook to eat better.
• Zara, a keen foodie. Bulk buys at an African store every two months and learns new dishes on YouTube.

The common thread: the food matters, but finding, planning and getting it is scattered across stores, social media, WhatsApp and phone calls.

Onboarding landed. Two people said the flags and regions made the app feel like theirs straight away. Cooking a recipe and ordering for a party were the hardest, and they were hard for the same reason.

What I found

People had to leave the app to finish what they started. Four of five testers hit it. To order from a caterer they had to phone them. To buy groceries they went to a shop’s website. To see a menu they left for a browser. Every exit was a place to drop off. Asked what would make them miss the app if it disappeared, testers named ordering in it.

Daniel said he would “fully switch” from Uber Eats if the caterers he orders from were in the app.

The other patterns:

• Add to list confused people (4 of 5). It pulled them off the recipe, added every ingredient instead of only what was missing, and never said which recipe an item was for.
• Recipes are the weekly habit. Testers ranked recipes first of the four parts (average 1.8 of 4), though every part was someone’s favourite.
• Stores weren’t sorted by distance (3 of 5), so people guessed which was closest.
• AI trust is fragile. Kola called its jollof “super close” to the real thing, but a bug that reset the chat and changed the recipe earned a 1 out of 5.
• Smaller asks (2 of 5 each): search by dish or ingredient, a servings control, swiping between steps with wet hands, and supermarkets as well as African grocers.

I fixed six bugs from these sessions within days, including a chat reset bug.

The redesign

Every change traces back to something testers said. The guiding rule: finish everything inside the app.

1. Order without leaving

Every caterer now has Place order: pick packages, choose delivery or pickup and a date, and follow the order live from sent to delivered. A Rootine Runner can shop and deliver groceries too, with a chat and a choice of what happens if something is out of stock. I chose this over linking out to delivery apps, which would have kept sending people away.

Before: Order on WhatsApp

After: Place order

2. A shopping list that knows why you need things

Adding to the list now stays on the recipe. Tapping an ingredient marks it Got it, so only what is missing goes on the list. Each item shows which recipe it is for, and the list is grouped by aisle.

Switch between the first build and the redesign to see the difference.

In the app, Get ingredients asks how you want them: shop it yourself from a list sorted by aisle, or let a Rootine Runner shop it, with every nearby store priced first.

Two ways to get it

The cheapest mix

3. Nearest first, and search everywhere

A postal code or your location sorts every store and caterer nearest first, with the distance on each card. One search on every tab finds dishes, ingredients (including other names, like dodo for plantain), stores and caterers. Supermarkets now sit next to African grocers, with prices.

The map grew with it. What started in Mississauga now covers the wider GTA: nine caterers from Oakville to Pickering, six Naomi African Food Market branches, and four supermarket chains.

4. Cooking that fits real kitchens

Servings chips rescale a recipe for one person or a crowd. In Cook Mode you swipe between steps, and the timer keeps running and notifies you even when you leave the app. And a recipe can come in from anywhere: paste a link from Instagram, TikTok, YouTube or a blog, or snap a cookbook page, and it arrives in the same format, ready for the list and Cook Mode.

5. AI that earns trust

The Rootine Plan asks five quick questions (cuisines, diet, allergies, daily targets, how long you want to cook) and builds a meal plan from tested recipes first. AI only fills the gaps. It came straight from watching one bug cost a tester’s trust.

Step 1: which cuisines

A plan day

6. Food is social

One tester asked for community, and it echoed the original research: people trust what their own people recommend. I added a light social layer to test the idea: post the dishes you cook, follow cooks, share lists with friends, and a spice rack that builds your taste profile from what you save. Games and trivia are parked.

Your kitchen and spice rack

A friend

Look and feel

The brand is warm and vibrant: deep forest green with a hand drawn food pattern, sunset orange for the main actions, goldenrod and cream, set in Space Grotesk and DM Sans. Flags and greetings from across the continent (Akwaaba, Karibu, Ẹ káàbọ̀, Bienvenue) make it feel like home from the first screen.

Every piece of small text meets WCAG AA contrast, 4.5:1 or better. I made darker shades of the orange and green for text on light backgrounds so the brand colours could stay.

App icon

Palette · tap to copy

• Deep Forest#00380E12.0:1 with cream
• Emerald#009A4C3.7:1 with white, decorative only
• Sunset Orange#FD9A326.3:1 with forest
• Goldenrod#F0C71D8.2:1 with forest
• Cream#F9F3D912.0:1 with forest

Made for small text

• #007038Emerald for text5.9:1 on the app’s cream, from Emerald at 3.5:1
• #A34C00Orange for text5.5:1 on the app’s cream, from Sunset Orange at 2.0:1

Space Grotesk · headlinesCraving that taste of home?

DM Sans · everything elseRecipes, grocers and caterers from across the continent, nearest first.

• AkwaabaTwi, Ghana
• KaribuSwahili, East Africa
• Ẹ káàbọ̀Yoruba, Nigeria
• BienvenueFrench, West and Central Africa

Feels like home

Splash

Onboarding

Home

Cook

Recipe

Cook Mode

Rootine Plan

Shop

List by aisle

How much?

Store

Order

Cart

Order tracker

On the way

Share

Community

A dish

A kitchen to follow

Motion

Motion explains what is happening. It never just decorates, and all of it respects Reduce Motion.

In real life

One short film shows the app where it gets used: on the kitchen counter.

Why ordering

People would pay for savings, not features. Four of five testers would pay for free delivery, lower fees or discounts, the way a food delivery membership works. None would pay for a recipe app alone. Most put the ceiling at about $10 to $11 a month. Funmi put it plainly: anything is too much “as soon as it hits two digits”.

That pointed the product at ordering, and one tester suggested caterers, not customers, should pay most of it. So the question became how much ordering to build first. I weighed four options.

Order requests won because they test real behaviour cheaply. The prototype shows what C and D would feel like end to end, so the idea can be tested with caterers before anything heavier is built.

An honest note: demand is still soft. Asked how disappointed they would be if the app disappeared, testers answered 2.5, 3 and 4 out of 5. The redesign targets exactly what they said would change that.

Where it stands

The redesign is built and in TestFlight. Ordering, Runners and the social features are a high fidelity simulation: they work end to end on the phone, with sample people and orders, so the ideas can be tested before a backend exists.

Early signal. I asked two people to try the redesign informally: a check, not a second round of testing. One called it multifunctional and singled out bringing in recipes from a link. The other summed up the whole redesign.

My normal experience is going through multiple food blogs, messy screenshots, the timer set up somewhere else. I like that it was all in one place.

She also liked finding grocers and caterers in the app, “another task eliminated”, and a place to share food: “when I make something I’m proud of, I send it around to the necessary people.” And she raised trust. Without a friend’s recommendation, Google reviews are how she decides on a caterer. That went straight onto the list.

What’s next

1. A second round of testing with the same script, so the numbers compare, and with East, North and Southern African testers.
2. Caterer reviews, ideally synced with Google reviews, for the moments a friend has not already vouched for one.
3. The ordering pilot with two or three caterers, with real payments.
4. A small backend for accounts, real orders and keeping the AI key off the phone, then the App Store.
5. One or two more cities, then South Asian and Middle Eastern communities, who share the same problem.

What I learned

The research was right the first time. Leaving ordering out of the first build was a fair call for a pilot, but five people told me in an hour each what 116 survey answers had already hinted. Next time I would test the riskiest assumption first, not the easiest one to build.

Building with AI changed how I work. I could put a real app in people’s hands in weeks instead of a clickable prototype, and watch where it broke. People were not imagining the app; they were using it. That made testing honest, and it made every design decision mine to defend, because every one of them shipped.

And trust is a design material. One bug cost a tester’s faith in the AI. One friend’s recommendation decides which caterer gets the call. Most of the redesign is about earning that trust, screen by screen.

1. Immigration, Refugees and Citizenship Canada: French speaking students from Africa, committee briefing, November 2024 ↗ (opens in a new tab)
2. Statistics Canada, The Daily, May 5, 2026: Estimating the international student population in Canada using administrative data ↗ (opens in a new tab)
3. Statistics Canada, The Daily, October 26, 2022: 2021 Census, immigration and ethnocultural diversity ↗ (opens in a new tab)