It is Sunday evening. You open the fridge, stare at half a cabbage and a jar of olives, and feel the familiar dread of the week’s most repeated question: what are we eating tonight? Multiply that by seven days, add a grocery run you keep postponing, and you have one of modern life’s quietest time thieves — meal planning.
Here is the good news: this is one of the tasks AI handles genuinely well. ChatGPT, Claude and their cousins are excellent at turning a few constraints (“two adults, one vegetarian teenager, €80 budget, shops at Lidl”) into a full week of dinners plus a shopping list. This guide shows you exactly how to plan weekly meals with AI, step by step, with copy-paste prompts and a worked example.
Why AI is surprisingly good at meal planning
Meal planning is a constraint-satisfaction puzzle: dietary needs, budgets, cooking time, seasonal produce, leftovers, variety. Humans are bad at holding all of that in their heads at once; language models are built for exactly this kind of juggling. Give an AI your rules and it will happily generate seven days of dinners, check them against your budget, and rework the plan when you say “no fish on Tuesdays”.
It is also tireless at the boring parts: converting a meal plan into a consolidated shopping list, scaling recipes up or down, and suggesting what to do with the half cabbage. Used well, it can cut your weekly food admin from over an hour to about fifteen minutes.
Step 1: Tell the AI who it is cooking for
The quality of your meal plan depends entirely on the quality of your brief. Before asking for a single recipe, write down the facts the AI needs:
- Household: how many people, ages, appetites (a hungry teenager eats differently from a toddler).
- Dietary needs: vegetarian, halal, lactose-free, gluten-free, allergies — be specific and non-negotiable here.
- Dislikes: the foods people simply will not eat. This is where most generic meal plans fail.
- Cooking capacity: how much time on weeknights versus weekends, your equipment (oven? air fryer? just a hob?).
- Goal: save money, eat healthier, reduce waste, try new cuisines — pick one primary goal.
Paste this into ChatGPT or Claude as your setup message:
“You are my weekly meal planner. Household: two adults and one vegetarian teenager in Amsterdam. No pork, no nuts (allergy). We dislike mushrooms. Weeknights: max 30 minutes cooking, we have an oven and a hob. Budget: €80–90 per week, we shop at Albert Heijn and the local market. Goal: healthy, varied dinners with minimal food waste. Acknowledge these constraints and wait for my next instruction.”
Starting with a setup message like this — instead of jumping straight to “give me a meal plan” — is the single biggest upgrade you can make. It turns the AI from a recipe slot machine into something closer to a personal nutrition assistant.
Step 2: Set your budget, time and shopping constraints
European grocery reality differs from the American recipe blogs most AI models trained on, so spell it out. Mention your actual shops — Lidl, Aldi, Carrefour, Tesco, Mercadona, Albert Heijn — because the AI knows their typical ranges and prices. Mention the season, because asparagus in May and root vegetables in January change what “cheap and fresh” means.
A useful follow-up prompt:
“It is late September. Which seasonal vegetables and fruits should form the backbone of our plan to keep costs down? Suggest 8–10 items commonly available and affordable in Dutch supermarkets right now.”
This grounds the plan in what is actually in the shops this week, rather than a generic international template. If you shop at discounters like Lidl or Aldi, say so explicitly — the AI will lean toward simple ingredient lists and fewer speciality items, which is exactly how you stay inside €80 a week.
Step 3: Generate the week
Now ask for the plan itself. Be demanding about format — a wall of text is useless on a phone in the supermarket:
“Create a 7-day dinner plan (Monday–Sunday) using the constraints above. Rules: each meal max 30 minutes on weeknights; at least two vegetarian meals; no ingredient used only once (avoid waste); total estimated cost within €80–90. Format as a table: Day | Meal | Time | Approx. cost. Add one sentence per meal explaining why it fits our constraints.”
Read the result critically. The AI will occasionally suggest something odd — a spice you do not own, a technique beyond a tired Tuesday. That is normal. Reply with corrections (“swap Thursday’s meal, we had lentils last week”) and it will revise. Two or three rounds of feedback is typical before the plan feels like yours.
Step 4: Turn the plan into a shopping list
This is where the magic compounds. Once the plan is settled, one prompt converts it into a single organised list:
“Convert this meal plan into a consolidated shopping list. Group items by supermarket section (produce, dairy, meat/fish, dry goods, frozen). Combine duplicate ingredients into single quantities. Mark with [HAVE] anything from this pantry list: olive oil, rice, pasta, tinned tomatoes, spices, flour, eggs.”
The pantry trick matters: without it, the AI will cheerfully tell you to buy olive oil every single week. Keep a running pantry list in your notes app and paste it in each Sunday — thirty seconds that saves real money.
Step 5: Get a prep-day game plan
If you do any Sunday prep, ask the AI to sequence it. This is an underrated use:
“I have 90 minutes on Sunday afternoon for meal prep. Based on this week’s plan, give me an ordered prep list: what to chop, cook and store, in the most efficient sequence, with storage notes (fridge vs freezer, how many days each keeps).”
Ninety focused minutes of prep — grains cooked, vegetables chopped, one sauce made — can take ten minutes off every weeknight dinner. The AI is genuinely good at spotting shared prep across meals, like chopping all the week’s onions at once.
A worked example: one full day
To make this concrete, here is what a day from an AI-generated plan might look like for our Amsterdam household (€85/week budget, 30-minute weeknights):
Tuesday: Turkish-style red lentil soup (mercimek çorbası) with warm flatbread
- Why it fits: fully vegetarian for the teenager, naturally nut-free, one pot, 25 minutes start to finish. Red lentils are among the cheapest proteins in any European supermarket.
- Ingredients: red lentils, one onion, carrot, tomato paste, cumin, paprika, lemon, flatbread from the bakery section.
- Approx. cost: €4–5 for three generous portions.
- Waste note: the lemon half goes into Wednesday’s salad dressing; leftover soup becomes Thursday’s lunch.
- Prep overlap: chop double onion and carrot on Sunday — half for this soup, half for Wednesday’s traybake.
Notice how each meal carries its justification, cost and waste plan. That is the format to demand from the AI: not just what to cook, but why it fits your life. When every meal earns its place on cost, time and taste, the plan survives contact with a real week.
Copy-paste prompt library
Save these somewhere handy and reuse them weekly:
- The reset: “New week, same constraints as last time. Generate 7 dinners with no repeats from the last two weeks.”
- The budget squeeze: “Money is tight this week: cap the plan at €60. Prioritise pulses, eggs, seasonal veg and frozen fish. Flag any meal over €8.”
- The variety injection: “We are bored of our usual rotation. Keep all constraints but introduce two cuisines we rarely cook (suggestions: Levantine, Korean-inspired). Keep spice requirements minimal.”
- The leftovers rescue: “I have leftover roast chicken, half a cabbage and 200g of feta. Suggest two meals that use all three, within our usual constraints.”
- The guest upgrade: “Friends are coming Friday — scale the plan to 5 people for that night only, and suggest one impressive-but-easy dish under 45 minutes.”
Common mistakes to avoid
- Skipping the setup message. Without constraints, you get a generic American meal plan with ingredients you cannot find. The five minutes of briefing pays for itself all week.
- Accepting the first draft. Treat the AI like a junior assistant: review, correct, iterate. The third version is usually the keeper.
- Ignoring portions. Tell the AI your household’s appetite explicitly, or a “serves 4” plan will leave a teenager raiding the fridge at 10pm.
- Forgetting food safety. The AI is not a food-safety authority. When in doubt about storage times or reheating rice, check an official food-safety source rather than trusting the chatbot.
Frequently asked questions
Which AI is best for meal planning?
ChatGPT and Claude both handle this well; Claude tends to follow complex formatting instructions slightly better, while ChatGPT’s memory features can remember your constraints between weeks. Either works — pick the one you already use.
Can AI handle allergies safely?
It can accommodate allergies in planning, but always double-check ingredient lists yourself, especially for severe allergies. AI suggestions are a starting point, not a medical guarantee.
How much time does this actually save?
Most people go from 60–90 minutes of weekly planning, list-making and recipe-hunting to about 15–20 minutes of prompting and reviewing. The bigger win is killing the daily “what’s for dinner” decision entirely.
Meal planning is just one of many weekly routines AI can quietly take over. If you enjoyed this kind of structured AI planning, you may also like our guide to creating a study plan with ChatGPT — the same constraint-first prompting technique, applied to learning. And for a different kind of planning project, see how to write a business plan with AI, which walks through a longer-form version of the same method.
2 thoughts on “How to Plan Your Weekly Meals with AI: A Step-by-Step Guide”