AI-first food app
Cooking is easy. Getting to cook is the hard part. The feature decision and the research was my major contribution.
Role- UX Designer (Final Year Project)
Breakfree Consulting Mumbai, India.
Project Timeline 2 weeks
I designed Platted — an AI-first food app
That bridges the gap between knowing what to eat and actually being able to eat it, without switching between three different apps to do so.
01
Why I built a fridge scanner instead of a recipe search bar.
A recipe search bar assumes you already know what you want to cook. My research showed users most often don't. The real moment of need is: "I have eggs, curd, and bread — now what?" A search bar is useless for that moment. A fridge scanner — where you photograph or manually list what's left — lets the app generate recipes from real constraints, not ideal-state ingredient lists.
Trade-off I accepted: Longer onboarding (2 extra screens). I reduced the exit risk by making the screens feel like personalisation, not a form.
Options considered
Rejected
Generic recipe search bar — type a dish name, see results. Fast to build, familiar pattern.
Chosen
Fridge scanner — photo or manual entry of what you have. App generates recipes from real inventory.
02
Why budget preferences live in onboarding, not settings.
My first instinct was to put budget as a filter — a settings page a user could optionally configure. But in testing I found that users who never set a budget filter got meal recommendations they couldn't afford, and they assumed the app wasn't for them. Moving budget into onboarding (cost per meal, meals per day, preference by meal type) made every recommendation contextually appropriate from session one.
Trade-off I accepted: Longer onboarding (2 extra screens). I reduced the exit risk by making the screens feel like personalisation, not a form.
Options considered
Rejected
Budget as a settings filter. Optional, hidden by default. Most users never touch settings.
Chosen
Budget in onboarding, framed as "personalise your plan." Set once, shapes every recommendation.
03
Why I added a daily check-in instead of a streak or score.
The original brief suggested gamification — streaks, rewards, leaderboards. But my research showed users felt guilt, not motivation, when they broke streaks. The "Intention to Action" check-in I designed asks: did you eat what you planned? How healthy was it? How do you feel? — no scoring, no failure state. It closes the loop without punishing the user and gives me data to improve future recommendations.
Trade-off I accepted: Less addictive than a streak system. I chose retention through usefulness over retention through anxiety.
Not a recipe app.
Not a grocery app.
The gap between them.
Every competitor solves one part of the food decision chain. Platted is the only design I found that connects all four.
Fridge-first recipes
Recipes generated from what you already own, not ideal-state ingredient lists. Reduces food waste. Makes cooking feel possible, not aspirational.
Zomato / Swiggy: order-first. Platted: pantry-first.
Budget-aware from day one
Every recommendation is filtered by your ₹/meal budget, set during onboarding. Healthy eating stops being a decision that costs more money.
HealthifyMe tracks. Platted decides within constraints.
Closed loop: plan → cook → reflect
The daily check-in feeds AI personalisation — your meal recommendations improve based on what actually worked for you, not generic nutrition data.
No competitor had a non-punitive feedback loop.
