MealPill

Problem

Immigrant families in Metro Vancouver plan meals across languages and stores. Food expires in the fridge because nobody tracks it, and the shopping list is whatever someone remembers on the way to the store.

Blueprint

Before. The week is planned in someone’s head. What’s in the fridge and when it expires lives nowhere. The same product has a different name in each language at home and in each store. The list gets written on the way to the store.

What the system decides:

1. What goes into this week’s plan, starting from what is already in the fridge and when it expires.
2. Which product is which across languages, so “молоко” and “milk” match one catalog item.
3. Nutrients and cost per plan, calculated by deterministic code from the catalog.
4. The shopping list that goes with the plan.
5. What the week looked like, in a weekly AI analytics summary.

After. One plan, one list, in the family’s language. Stripe sits in front: trial, paywall, subscription, Customer Portal, cancellation.

What I built

• A web app (PWA): weekly meal plan from fridge contents, expiry dates, shopping list, weekly AI analytics.
• Localization in English, Ukrainian and Russian, end to end.
• A product catalog with cross-language matching and a deterministic nutrient and cost calculation.
• Stripe in live mode: subscription, paywall, Customer Portal, cancellation.
• An attribution funnel from first touch to subscription, plus AI token cost measurement.
• Trilingual Terms and Privacy set up, a consent guard, BPCPA and CASL compliance.
• QA on my own: price and matching audits, a P0 to P3 bug backlog.

Pricing: CA$8.99 a month or CA$89.99 a year, 7-day trial, card required, no free tier.

Results

Payments are live. No usage numbers until there are some worth showing.

Lessons

• Nutrients and cost come from deterministic calculation over the catalog, so the same plan always gives the same numbers.