Skip to content
AITellar
Get the Free Library Library

AI Meal Planning for Beginners: Your First Week, Step by Step

Meal Planning Updated 6 min read

An open notebook with blank ruled pages on a kitchen table in morning light, a pen resting in the gutter and a mug of tea beside it
What’s in this guide

The first week is the only hard one, and it is hard for a reason worth knowing before you start: you are describing your household to something that has never met it. That description is most of the work, and you only do it once.

Budget 25 minutes. Week two runs closer to 15, and by week four there is usually nothing left to correct. Nothing here needs a paid plan, an app, or a list of what is in your cupboards.

If you would rather have the stopwatch version than the reasoning, the 15-minute weekly method is the same job with the explanations taken out. This page is for the first time you do it.

What each step of the first week prevents Four steps, each paired with the first-week mistake it prevents. One, count the nights you actually cook, which prevents a seven-night plan that fails on Wednesday. Two, name what you will not eat, which prevents a week you have to redo. Three, correct once in a single message, which prevents three rounds that never converge. Four, save the filled-in prompt, which prevents starting from nothing next Sunday. Step four is marked as the one that decides whether there is a week two. THE STEP THE FIRST-WEEK MISTAKE IT PREVENTS 1 Count real nights Not seven. The ones you cook A plan that fails on Wednesday 2 Name the no list Allergies, and what starts a row A week you have to redo 3 Correct once Everything wrong, one message Three rounds that never settle 4 Save the prompt A note on your phone will do Starting from nothing next week Week one is about 25 minutes. Only step four decides whether there is a week two
Each step exists to prevent one specific first-week failure. Step four is the one that decides whether there is a week two at all.

What you need before you start

Open ChatGPT. The free tier is fine and signed-out works. Have somewhere to put the finished plan — a note on your phone, a bit of paper on the fridge, the printable at the end of this page.

That is the whole setup. You do not need an inventory of your kitchen, a diet to follow, or any idea of what you want to eat. Not knowing what you want to eat is the problem you came here with, and it is the one thing this method does not ask you to solve first.

Step one — count the nights you actually cook

Not seven. Count the evenings you genuinely stand at the hob, and leave out the one that is always a takeaway, the one somebody else handles, and the one you eat at your mother’s.

Most households who think they cook seven nights cook four or five. A plan for five nights that survives is worth more than a plan for seven that collapses on Wednesday and takes your confidence in the whole idea with it.

Step two — describe the household once

This is the part that takes the longest the first time and no time ever again. Paste this in and fill in the brackets, leaving the brackets themselves in place. Every line in it is there for a reason, and the prompt taken apart line by line says what each one prevents.

The beginner's first-week prompt
I have never planned meals this way before. Help me plan dinners for one week. People: [how many adults and children, and rough ages] Nights I actually cook: [the number, and which ones if it matters] Time on a normal night: [how many minutes] We do not eat: [allergies, dislikes, anything that starts an argument] Kitchen confidence: [beginner / fine / confident] Budget: [tight / normal / not a concern] Give me one dinner for each night I cook. One line each: the meal, roughly how long it takes, and one sentence on why it fits my household. Rules, all of them and not most of them: 1. Nothing that needs an ingredient from a specialist shop. 2. No main protein appears twice in the week. 3. At least two need to be doable in under 25 minutes. 4. Nothing on the do-not-eat line. Check your list against rules 1 to 4 before you reply, and fix anything that fails. Then stop. I will tell you what to change before you write a shopping list.

The kitchen confidence line is the one beginners leave out and the one that changes the answer most. Say beginner if you are one. It is not an admission, it is a constraint, and constraints are the only thing that make the reply useful.

Step three — read it, then say everything wrong at once

What comes back is a draft, and reacting to a draft is a far smaller job than inventing five dinners from nothing. That difference is the entire reason to do it this way.

Read all of them before you judge any of them

Going line by line makes you negotiate with each dinner separately, and turns five minutes into twenty. Read the whole list, then decide.

One message, everything in it

Not three messages. Something like: Nothing with fish. Tuesday has to be under 20 minutes. Swap the risotto, nobody here will eat it.

It rewrites the whole week accounting for all three at once. Fixing them one at a time produces a patched plan, and the patches are where the bad dinners live.

Two rounds, then stop

A third round of corrections moves the disappointment rather than removing it. If two dinners are still wrong after two rounds, keep the three that work and decide the others on the night — which is what you were doing for all five before you started.

Step four — save the prompt, and why this is the real step

Copy your filled-in prompt into a note on your phone before you close the tab.

Everything above this line was week one. This step is what makes week two 15 minutes instead of 25, because the describing — the part that actually took the time — is now written down and reusable. Skip it and you will do the whole first week again next Sunday, decide the method is slow, and stop.

Add the corrections too. Whatever you had to fix this week, paste into next week’s prompt, and the first pass arrives closer to right. After a month there is usually nothing to correct. The tool has not improved. Your description has.

Turning the plan into a shopping list

Once you have approved a week, get the thing you actually carry to the shop.

The shopping list follow-up
Turn that plan into a shopping list. Group it by supermarket aisle: produce, dairy, meat and fish, dry goods, freezer, other. Quantities for [how many people], written the way a shop sells them — a bunch, a tin, 500g — not cups. Put a star next to anything used in only one meal. At the bottom, a short CHECK FIRST list of things I probably already have.

The starred items are where a sensible week quietly turns expensive — each one is bought for a single meal. Sometimes the answer is buy it anyway. The star makes it a decision instead of a surprise at the till. There is more on this in the grocery list prompt.

What goes wrong in week three

Not week one. Week one has novelty carrying it. Week three is when something goes wrong in the week itself, the plan gets abandoned on Tuesday, and the whole idea feels like it failed.

It did not fail. A week got away from you, which happened before you had a plan as well. The mistake is restarting from nothing the following Sunday. Open the last plan that worked, paste it back in, ask for a week like that one, and you are running again in about three minutes.

Keep a running list of the dinners that actually got eaten. After two months that list is worth more than any prompt on this site, because it is specific to your household and nothing else can produce it. The full meal planning guide covers what to do with it once it is long enough.

The printable

The Weekly Dinner Planner is one sheet: seven lines for the dinners, a column for what needs buying, and the prompt cards on the back so you are not looking this page up again next Sunday.

Questions about starting out

Do I need to pay for ChatGPT to do this?

No. Everything on this page runs on the free tier, and neither prompt uses a paid feature. Paid plans are faster and remember more between sessions, which is convenient rather than necessary. Nothing in the method depends on it.

How long does the first week really take?

About 25 minutes, and roughly 15 of those are describing your household for the first time. That description is a one-off. Week two is closer to 15 minutes total because you are pasting a saved prompt rather than writing one.

What if I do not know how many nights I cook?

Count last week rather than a normal week. Normal weeks are what you intend; last week is what happened. If last week was unusual, count the week before. Being one night out matters much less than guessing seven.

Do I have to tell it what is in my cupboards?

No, and a full inventory would waste the time you are trying to save. Three or four things you want used up is plenty — the half bag of spinach, the mince in the freezer. It will build around them and assume the rest.

What if everything it suggests is wrong?

That usually means a constraint is missing rather than that the method failed. Add the thing you are reacting to — too much pasta, too fussy, too many pans — as an explicit rule and ask again. Two rounds of that is normally enough.

Keep going