Food & cooking

What food and cooking apps are worth building?

Recipe collections are plentiful, so the openings sit in the gaps between recipes: the moment after the shop, the date nobody checked, and the diet rule a generic label does not answer. Ideas that begin with a physical input, such as a photo of a fridge or a barcode, are easier to explain than ideas that begin with a blank search box.

• Inventory to meal: the user lists or photographs what is in the house, and the app suggests two or three meals that use it first.
• Waste reduction: the app ranks what to cook first by how soon each item expires, and says why.
• Dietary checks: the app confirms whether a product meets a rule such as halal, kosher, vegetarian or a named allergy, using the ingredient list the user provides.
• Batch planning: one cook for three days, with a shopping list that matches the batch.

Each of these depends on data the user can actually give you. If the first version has to work on day one, a typed list beats a photo the app has to read.

Who cooks with an app, and how do you find them?

The core users cook for other people: a parent deciding dinner for four, someone cooking for a partner with a coeliac diagnosis, or a shared flat where three people buy food separately. They can use a tool every day, but recipes are free almost everywhere, so charging for them is a hard test. A feature that saves money on food is a better one to test for payment.

Reach them where food problems are discussed: meal-planning forums, community groups for specific diets, and threads where people share tips on cooking leftovers. Ask the people there what they threw away last week. That answer is more useful than any survey you could run.

• Post a one-page version in a diet-specific community and ask people to run it against their own fridge.
• Count how many people who tried it asked for the list at the end. A saved list is a stronger sign than a like.
• Keep a note of every food that failed to match. Those misses show the gaps in your data.

What makes a food idea good?

• A repeating moment: the question comes up every day or every shop, not once.
• Input the user already has: a photo, a receipt or a list in a notes app. Do not ask for a new habit.
• An answer that can be checked: the recipe, ingredient list or expiry date is visible, so the user can overrule the app.
• Honest limits: the app says what its data covers and where it could be wrong.
• A reason to come back: leftovers planned for tomorrow give people a reason to open it again.

What would a first kitchen app look like?

Hypothetical example: a household of two has a fridge full of items close to their dates. The first version takes a typed list, asks how many people are eating and how much time there is, and returns three dinners. Each dinner shows which listed items it uses up and which are still left over.

Build a single-page dinner suggester. Input: a typed list of ingredients, the number of people, and the maximum cooking time. Output: three meals, each listing the ingredients it uses from the list and the items left over. No sign-up. Do not invent quantities the user did not give. Mark any ingredient the user must buy as optional.

A first build fails in two recognisable ways. Either people type three items and leave, which is an input problem, or the suggested meals ignore what they listed, which is a quality problem. Each needs a different fix, so watch which one happens before changing anything.

What are the risks in food and cooking apps?

• Allergies and medical diets. A wrong answer about nuts or gluten can cause real harm. Do not present the app as a safety check for allergens, and keep a clear warning beside every allergen answer.
• Religious certification. A label is not the same as a certification. Halal and kosher rules and the bodies that certify them differ by region, and a may-contain note needs care. Say what the app checked and what it could not.
• Food date rules. In the UK and the EU, a use-by date relates to safety and a best-before date relates to quality. Get that wrong in your copy and people will act on the mistake.
• Product data. Barcode and ingredient databases are often licensed or incomplete. Read the terms of any source before you rely on it.
• Photos of kitchens and receipts. These show personal information, including family members and addresses. Delete images after processing unless the user keeps them.

Which food ideas are worth testing first?

Four food ideas are on the board at the time of writing, and two of them describe the same fridge-photo job. Fridge photo to three dinners matches the worked example above, and a photo of the fridge, and dinner is decided describes it again from a different angle. Cook what’s about to go off, not what’s in the recipe is the waste job, with output ranked by expiry. Yuka for halal and kosher is the dietary-check job, and it is the one to read the risks section for before anything else.

1. Keep a three-day log of what you threw away and why.
2. Build the dinner suggester from the prompt above, with typed input only.
3. Look through the food and cooking ideas. If your kitchen problem is not there, post it with the input a cook would actually have to hand.
4. Read how to write an app build prompt before you give the brief to a coding tool, and how to scope an MVP to keep it to one screen.