“What do I cook today?” is probably the most repeated question in any home. AI won’t cook for you, but it solves surprisingly well the part that takes the most willpower: deciding the menu, using up what’s in the fridge and building the shopping list. We’ve been using it for this for months, and here’s the honest guide: what genuinely works, the exact prompts and the limits worth knowing.
What AI is genuinely good for in the kitchen
- Planning weekly menus: give it your conditions (people, budget, cooking time, tastes) and it builds the whole week in a minute.
- Emptying the fridge with judgement: tell it what you have and it suggests what to cook with it, instead of letting food spoil.
- Adapting recipes: portions, missing ingredients, no-oven version, lighter version. Its strongest point.
- Automatic shopping list: from the weekly menu to a list organised by supermarket sections, in one step.
The prompts we actually use
- Weekly menu: “Make a lunch and dinner menu for 2 adults and a child, 7 days, max 30 minutes per dish, simple everyday cooking, tight budget. Reuse ingredients across days to avoid waste.”
- Fridge: “I have: [list what’s there]. Suggest 3 different dinners using only this plus pantry basics. Say how long each one takes.”
- Adaptation: “Adapt this recipe to 6 people and gluten-free: [recipe]. Bold what changes.”
- Shopping list: “From the menu above, make me a shopping list grouped by sections: produce, meat, fish, pantry, dairy.”
Our experience: what works and what doesn’t
- What changed our week the most: the menu + shopping list combo. Fifteen minutes on Sunday and the daily “what do I cook?” is gone. One of the best effort-to-benefit AI uses we know.
- The trick that makes the difference: real context. “Make me a menu” returns something generic; telling it who eats, how much time you have and what you hate returns something usable on the first try.
- Where it fails: it sometimes suggests odd combinations or unrealistic quantities, and “invented” recipes can have confusing steps. For specific techniques (doughs, fermentation, doneness), a good book or a cook still wins.
- The mistake to avoid: following it blindly. Read it like you’d read a clever friend’s suggestion: almost always useful, not infallible.
Watch out for this (important)
- Allergies and intolerances: AI can get hidden ingredients wrong (traces, derivatives). If there’s a serious allergy at home, always verify labels and reliable sources; don’t delegate that to a chatbot.
- Medical diets: for health conditions (diabetes, kidney disease, pregnancy…), AI doesn’t replace your doctor or dietitian. Use it to vary dishes within the guidelines you’ve been given, not to create the guidelines.
- Food safety: storage times and risky cooking (chicken, eggs, home canning) are better confirmed with official sources.
Frequently asked questions
Which AI should I use, do I need something special?
Any general assistant (ChatGPT, Gemini, Claude) does it well; you don’t need a dedicated cooking app. If you want to do it from your phone while shopping, see the best AI apps for your phone.
Are AI-invented recipes reliable?
For everyday dishes, generally yes. For delicate techniques or exact baking quantities, cross-check with a tested recipe: that’s where it gets things wrong the most.
Conclusion
AI won’t make you a better cook, but it takes away the worst part: deciding, planning and making the list. Start with Sunday’s menu using our prompts and adjust them to your home. If you like organising your life this way, the natural next step is to plan your whole week with AI.
