Skip to main content

Generate a Meal Plan with AI

Generate a Meal Plan with AI Why: Turn a plain-language idea into a meal plan draft using recipes and Bites you already have, so you can spend less...

Written by Ken Truesdale

Platform 2 Guide — This article applies to the new platform. Looking for classic platform docs? Browse the Platform 1 collection.

Why: Turn a plain-language idea into a meal plan draft using recipes and Bites you already have, so you can spend less time arranging meals and more time creating content for your members.

Navigate: App Station > Content Library > Add Content > Generate Meal Plan

Quick Start

  1. Select a meal plan template.

  2. Enter a description of the plan you want.

  3. Review the interpreted settings and edit them as needed, including Match words.

  4. Select Generate.

  5. Open the draft in the meal plan editor and make any changes.


What the Meal Plan Generator does

Describe the meal plan you want in plain language, and the Meal Plan Generator creates a draft using recipes and Bites from your existing library. It considers recipe titles, ingredients, tags, and member engagement, such as favorites, ratings, and how often recipes have been cooked.

The generator does not invent recipes, cooking steps, images, or member-facing copy. After generation, you can edit the draft in the regular meal plan editor. Generation is a one-time step; you can’t regenerate or use AI to edit the plan from inside the editor.

Note: Recipe text sent for generation is not used to train the model.

Who can use it

The Meal Plan Generator is available to App Station admins on the Growth plan. Growth includes 10 AI calls per hour. Reviewing settings and selecting Generate each count as one call.

Member Kitchens can raise the hourly limit for your app. A blank limit means there is no hourly cap. You can’t change this limit from your kitchen settings.

To open the generator, go to Content Library > Add Content > Generate Meal Plan. To manage the rules that apply to every generated plan, go to Features > Meal Plans > Generator.

Set your standing generator rules

Standing rules apply to every plan you generate. They can narrow the recipes or settings available for a particular request.

Go to Features > Meal Plans > Generator to review or update these rules:

  • Eligible collections: Limit recipes to selected collections. Leave this empty to use the whole library.

  • Excluded tags: Keep recipes with selected tags out of generated plans.

  • Required dietary tags: Require recipes to have selected dietary tags.

  • Allergen exclusions: Exclude recipes containing selected allergens.

  • Section-to-meal-type map: Match meal plan sections, such as breakfast or dinner, to the meal types used to tag recipes.

  • Allow Bites for snacks: When enabled, Bites can be used in snack sections.

  • Daily nutrient targets: Set minimums and maximums for nutrients your app displays. Add only nutrients displayed in your app; calories and protein are included when you add them. Each target applies to the day’s total after servings are counted. Recipes missing a targeted nutrient may be less likely to be selected.

  • No-repeat window: Set how long recipes should be kept from repeating.

  • Include unpublished recipes: Choose whether unpublished recipes can be considered.

  • Default servings: Set the serving count used when a prompt doesn’t specify servings.

  • Leftovers: Leftovers are on by default. You can change this in the standing rules.

  • Maximum overlap with a recent plan: Limit how many recipes a new plan can share with any one recent admin plan. The default is 40%.

  • History window: Set how many days of recent admin plans are considered when checking for variety.

  • Use member favorites and ratings: Choose whether member engagement helps guide recipe selection. This is on by default.

  • Philosophy notes: Add optional guidance for the AI, such as your approach to meal planning.

There is no maximum cook-time setting. Recipe cook times may help guide selection, but they are not used as a hard filter.

Tip: Keep dietary and allergen rules accurate, and make sure recipes have the right tags and nutrition information. This gives the generator better information to work with.

Generate a meal plan

Choose a template and describe your plan

Select a meal plan template to set the plan’s layout, then describe what you want in the prompt. You can mention the number of days, meals, servings, dietary preferences, or leftovers.

The generator uses your prompt to suggest settings for review. If AI is unavailable, you can still fill in the review form by hand and generate a plan using the local assembler.

Review the settings

Check the interpreted settings before generating. Depending on the template, you may be able to review or change the title, number of days, sections, tags, nutrient targets, servings, leftovers, and variety notes.

Match words is a comma-separated list of words the generator uses to find recipes that fit your request. For a loose prompt such as “cozy fall,” review and edit the suggested words so they reflect the recipes you want. You can also enter Match words yourself if AI is unavailable.

Keep or remove empty days and sections

Keep empty days and sections is on by default. This keeps days and sections from the selected template in the draft even when the generator doesn’t fill them.

Turn it off to save only the days and sections selected on the review step. For a days-only layout, the review form hides section and leftover settings, and the generator places one recipe on each day. A simple list layout has no days or sections; choose how many recipes to place, and they’ll be saved in order.

Generate and edit the draft

Select Generate to create the draft. When it’s ready, open it in the meal plan editor to review the meals and make any changes.

The plan is saved as a draft and is not visible to members until you publish it.

How the generator handles variety

The generator learns from your recent admin meal plans to make new plans feel familiar while still offering variety. By default, a new plan can share up to 40% of its recipes with any one recent plan. You can adjust this overlap limit and the history window in your standing rules.

The generator may also use member favorites, ratings, and cooking activity to help choose recipes when that setting is enabled. It uses engagement summaries, not member identities or rating comments.

Leftovers, servings, and nutrition in prompts

Leftovers

When leftovers are enabled, the generator uses a source meal—usually dinner, or the section mapped as the source—to make extra servings. It then places a leftover session in the mapped section on the following day. The final day has no leftover session because there is no next day.

If your prompt asks for an exact number, such as “exactly 3 leftovers” or “exactly 3 lunches should use leftovers,” that count appears in the review settings and the plan places that many leftover meals. Leave the count blank to repeat leftovers every day after the first.

Servings

A serving count in the prompt replaces your standing Default servings setting. For example, “serves 2” or “for one adult” sets the serving count for the plan.

If a serving count only describes leftover dinners—for example, “those dinners should be 2 servings”—it won’t replace the plan’s serving count when the prompt also says who the plan is for.

Nutrient targets

Qualitative phrases such as “low fat” or “high protein” fill in the same daily nutrient targets each time. A specific number in the prompt, such as “at least 150g protein,” “under 40g fat,” or “1800 calories,” is treated as that day’s total.

Standing nutrient rules can still make a target stricter: they can raise a minimum or lower a maximum. Nutrient targets apply to the day’s total after servings are counted.

What to expect from a generated plan

Some meal slots may remain empty if your library is too small or recipes are missing nutrition information needed for the targets you set. Review the draft and fill any gaps in the meal plan editor.

The generation summary shows whether AI composed the plan, how many slots were repaired or filled in afterward, and the closest existing plan. These details can help you understand how the draft was assembled and whether you want to adjust it.

Did this answer your question?