How to Find Recipes from the Ingredients You Already Have (An Honest Guide)

You open the fridge. There are four things on the shelf and one bunch of something wilting in the drawer. You don’t want to order delivery. You don’t want to go to the store. You want a recipe that uses what you have, that you can start cooking in ten minutes, and that you will actually enjoy eating.
This is a question people have been solving the same five ways for about fifty years. Here’s the honest tour, ranked from oldest to newest, with the failure modes attached.
1. Ask a person who cooks
Text your mom, your grandmother, the one friend who is genuinely good at this. Tell them what you have. Wait.
This works astonishingly well when the person is available. The failure mode is that they are not always available, and that they will sometimes give you a recipe that requires three things you forgot to mention you don’t have.
2. Google the ingredients with “recipe” appended
The default move. “Chicken broccoli rice recipe” returns 200 million results. You click the first three. Two of them have a ten-paragraph backstory before the recipe. One of them wants you to subscribe to a newsletter to see the ingredient list.
The failure mode here isn’t the recipes themselves. It’s the funnel they sit in. Modern food blogs are optimized for ad impressions, not for someone who needs an answer in twenty seconds. You will get a recipe eventually but the path is deliberately slow.
3. Ask ChatGPT or Claude
Newer move. “I have chicken, rice, and broccoli. Give me a recipe.” You will get a competent-looking recipe in fifteen seconds.
The failure mode is the one we keep coming back to: the model is rewarded during training for producing plausible recipes, not for producing recipes that only use what you have. So it will add garlic. It will add “a splash of soy sauce.” It will add “1 tablespoon of cornstarch.” You will either skip those (and degrade the dish) or have a half-cooked chicken stuck in a pan while you check whether you have cornstarch.
We wrote a longer piece on why generative AI hallucinates ingredients if you want the full version.
4. Use SuperCook or a similar pantry-input app
Type all your ingredients into SuperCook. Toggle “pantry staples” on. You get a list of recipes that match.
This works well for people with deep pantries. The failure modes are: (1) you have to type or click thirty things, which most people give up on by ingredient eight, and (2) the recipes are scraped from the open web, so when you click through, you’re back in the food-blog ad funnel from option 2.
5. Snap your fridge with a camera app like CookSnap
Newest approach. Open the recipe finder and type a few ingredients, or open the iOS app and take a photo of your fridge. Computer vision pulls the ingredient list out automatically. You get one or two real recipe matches from a curated library, with a fit percentage and a list of any missing items.
The failure mode — we’re being honest — is that our library is smaller than the open web. If you have an unusual ingredient and demand a perfect match, the matcher will sometimes return “closest options” rather than a 100% hit. We tell you when that happens rather than papering over it.
Which one should you actually use?
Here’s our honest recommendation, by situation.
- You have a deep pantry and time to scroll.SuperCook will surface more options than anyone else.
- You want a creative idea, not a tonight’s dinner.Generative AI apps (DishGen, ChefGPT) are fine for “give me a new way to use this leftover cauliflower.” Just don’t cook the literal output without reading carefully.
- You have four ingredients and twenty minutes.That’s the case CookSnap was built for.
- You have all night and curiosity. Text the best cook you know. Nothing beats that.
The deeper point
Every approach above is solving the same problem with a different trade-off. There is no universally correct answer. The honest question is: at 7:30pm with hungry people in the kitchen, which tool gives you the fewest decisions to make?
We think it’s ours. We’ll let you decide.