RecipeFit, 2024
Recipes that start from what's in your fridge
Search by ingredients, allergies, and macro goals, with AI substitutions when you're missing something. Validated with 16 interviews and a 45-person survey.
Between tight schedules, missing ingredients, allergies, and fitness goals, cooking a healthy meal can feel like solving a puzzle. For a lot of students it means bouncing between five apps just to make dinner.
RecipeFit set out to make it as easy as possible to find a recipe that works for you, with what you already have. I was the product manager on a team of five, responsible for feature scoping, user validation, and our AI integration strategy.
Discovery
We interviewed 16 students and early-career professionals who cook at least twice a week. Four frustrations came up again and again:
- Ingredient mismatch. “I don't have cumin or coconut milk, so I just give up.”
- Allergy anxiety. “Even if a recipe says peanut-free, I still triple-check.”
- Macro confusion. “I want something high-protein, but I'm not calculating grams manually.”
- Recipe fatigue. “I scroll for ten minutes before realizing I don't want any of it.”
A broader survey of 45 people backed this up: 82% search for recipes based on what's in their fridge, 68% want to track some form of nutrition, and 60% had tried and abandoned a recipe app.
What it meant for the product
| Insight | Decision |
|---|---|
| People start with ingredients, not recipes | Ingredient-first search |
| Allergies are high-risk | Safety filters front and center, and strict |
| Users want to feel healthy, not micromanage | Show nutrition clearly without overwhelming |
Existing tools each did part of the job. Yummly had great search but weak macro and allergy filtering. MyFitnessPal tracked nutrition well but wasn't built for discovery. Tasty and Pinterest were inspiring but not goal-oriented. None of them let you say: here's what I have, here's how I eat, what can I make?
What we built
- Recipe finder. Enter ingredients, allergies, macros, and preferences like “quick meals” to get curated results.
- Substitutions. Missing something? Gemini suggests a swap that keeps nutrition and flavor balanced.
- Nutrition cards. A protein, fat, and carb breakdown on every recipe, flagged when it matches your goals.
- Profiles. Set your filters once, save favorites, and see what you've made before.
The stack was React and Tailwind on the front end, Node, Express, and MongoDB on the back end, and a small Flask service for Gemini. Edamam handled nutrition and allergy tagging.
Hard calls
Specific vs. flexible
Too many filters returned nothing; too few returned junk. We let users mark filters as must-have or nice-to-have.
Taming substitutions
Early prompts suggested things like replacing chicken with applesauce. We restructured them to include recipe context, diet labels, and formatting rules.
Earning allergy trust
If a recipe said “may contain peanuts,” we excluded it from peanut-free results entirely. Letting one unsafe recipe through costs far more than hiding a safe one.
Outcome
We shipped a working full-stack MVP in five days. In testing with 10 users after launch, 80% said they'd use it over searching Google.
Next on the list: fuzzy ingredient matching (your fridge has tomato sauce, the recipe wants crushed tomatoes), community tags like “dorm-friendly,” and a weekly meal-planning mode.