NutriAgenda, AI Meal Planner

November 4, 2026

Deciding what to eat involves more than choosing a recipe. There is what you like, what is already in the kitchen, how much time you have, and who else is eating with you.

NutriAgenda is a mobile app that brings those decisions together. Nutri, its AI assistant, helps create a meal plan from your preferences and everyday constraints. The plan then connects to recipes, nutritional information, a pantry and a shopping list.

The aim is to make a plan you can actually follow, from deciding to cooking.

NutriAgenda’s English welcome screen beside the Home screen showing meals for the selected day — The iOS app, captured in the simulator. Screens throughout this post use demonstration data; nutritional values are illustrative.

Start with your food preferences

Onboarding asks about goals, activity, food preferences and restrictions, one question at a time. These answers give Nutri context before it proposes meals.

Food preferences are more expressive than a yes or no. Tap an ingredient to say you like it; tap again to mark it as a favourite. Restrictions are separate, with a severity setting where relevant.

What someone enjoys and what they need to avoid are different inputs. Keeping them separate makes the choices easier to understand.

English onboarding with food choices grouped into fruit, vegetables, proteins and grains, including selected favourites — Choose foods you enjoy, then record dietary restrictions separately. These selections are sample answers.

Dietary restrictions step with lactose-free selected and a severity control — Choose foods you enjoy, then record dietary restrictions separately. These selections are sample answers.

The setup can also include a household. Other people can be added by name, with their own preferences and restrictions, without each person needing an account. When creating a plan, you can choose which household members to include.

Learn the flow by doing it

The questionnaire leads directly into a guided first plan. Instead of finishing setup and leaving someone at an unfamiliar dashboard, the tour walks through the next useful actions: create a plan, add its ingredients to shopping, activate it, and try marking a meal as completed.

The recording below follows the full journey, including the optional household questions. Use the chapters to move between the questionnaire and the tour.

The complete onboarding and first-plan tour in the iOS simulator. A fictional profile is prefilled; account creation, generation and shopping responses are simulated. Pauses are edited.

The final step uses a disposable meal preview. Trying the completion action does not change the real plan’s progress. A small celebration then marks the end of the tour.

Turn preferences into a plan

Creating a plan starts with practical choices: a week, a fortnight or a month; the number of meals per day; and when to prepare leafy vegetables. An optional budget is entered in meticais. Special requests add context such as quick dinners or familiar ingredients.

Generation runs as a background job with progress stages, so the app can show what is happening while the plan is prepared. It selects recipes from the catalogue and arranges them around the profile and planning choices.

Generate with Nutri form in English with weekly duration, three meals per day, a sample budget and special requests — Define the shape of the plan, then browse the result by day. The cookbook cover gives each plan a recognisable identity.

A sample active weekly plan with a cookbook cover, day selector and meal list — Define the shape of the plan, then browse the result by day. The cookbook cover gives each plan a recognisable identity.

AI generation is one route into a plan. The app also supports building one manually and exploring plans shared by the community. Importing from another service is marked as coming soon in the version shown here.

How the generator builds a plan

Generation combines AI assistance with backend planning and validation. Nutri uses food preferences and requests to help assess recipes from the catalogue. The backend organises the meals into a calendar, handles household portions and checks the plan against dietary restrictions and other planning requirements. These checks do not certify allergy safety.

The resulting plan connects to recipe ingredients and the shopping list. Household size changes the quantities needed; pantry stock changes what remains to buy. Budget estimates show their limitations when price data is incomplete.

Explore those relationships below. Change a favourite food, adjust the servings, then check what is needed for shopping. This simplified example illustrates the effects of the inputs.

EXPLORE A SAMPLE PLAN

From recipes to a calendar

See how preferences, portions and pantry stock affect a plan.

Favourite food

Servings per meal

YOUR CONTEXT

Preferences, dietary restrictions and household size give the planner context.

Favourite foodFish

Servings per meal2

Dietary restrictionNo peanuts

1 of 4

Interactive illustration with fixed sample menus, quantities and prices. No AI request is made; this example does not reproduce the production algorithm.

Bring the day into focus

A weekly plan is useful for looking ahead. Home answers a smaller question: what am I eating today? Choose a date to see its meals and a summary of the available nutritional values.

Meals can be marked as completed or skipped. Skipping has a reason, such as eating out, having no time or missing ingredients.

The plan detail keeps the wider schedule nearby. Switch days, open a meal, or browse its animated cookbook cover to look through the recipes.

Talk through an idea with Nutri

In the chat with Nutri, ask for a quick dinner using ingredients at home, or talk through a different meal idea.

Responses arrive as they are written. The composer supports text, dictation and attachments, so a question can include a photo or document as well as a message.

English chat with Nutri showing a fictional question about rice, broccoli and eggs, followed by a sample dinner suggestion — A sample conversation in the real chat interface. This response is demonstration content.

Find a recipe worth opening

The recipe list brings together meals from the active plan, personal recipes and the recipe catalogue. A staggered card grid gives each recipe its own colour, photo and preparation-time label.

Search helps narrow the list. A long press opens actions such as favouriting a recipe or inspecting its nutrients, keeping those details accessible without adding buttons to every card.

The English Recipes tab with a staggered grid of coloured recipe cards, food photos and preparation times — Recipes remain browsable as a collection, alongside the pantry and shopping list. Sample recipe data is shown.

Open a recipe and start cooking

A meal opens into a recipe with a photo, preparation time, servings and energy information when available. Ingredient chips keep names and quantities together; numbered instructions explain the preparation.

Cooking mode changes the pace. It presents one step at a time, with the ingredients for that step and controls to move forward or revisit an earlier instruction. A kitchen timer is available while cooking.

Lemon fish with rice sample recipe with a photo, duration, servings, ingredient quantities and numbered instructions — Read the recipe as a whole, then follow it one instruction at a time. Sample recipe content is shown in the real app interface.

Cooking mode showing the first instruction, fish and lemon quantities, step navigation and a timer control — Read the recipe as a whole, then follow it one instruction at a time. Sample recipe content is shown in the real app interface.

Know what the numbers describe

Nutritional information is available at two useful levels: the meal total and an individual ingredient’s quantity in the recipe. Opening an ingredient explains its contribution without leaving the recipe.

Meal nutrients sheet showing the sample meal total, energy, protein, carbohydrates, fat, fibre and sodium — The meal total and the values for 200 g of one ingredient describe different quantities. The app states the basis above the numbers.

Fish ingredient sheet showing the nutritional values for 200 grams in the sample recipe — The meal total and the values for 200 g of one ingredient describe different quantities. The app states the basis above the numbers.

A nutritional value needs a basis: per reference quantity, ingredient amount or whole meal. NutriAgenda also distinguishes approximate, partial and unavailable information. Missing data should not look like zero.

Try that distinction below. Select an ingredient to see its sample values, then return to the whole meal.

EXPLORE THE RECIPE

What makes up a meal?

Choose an ingredient to inspect its contribution.

Lemon fish with rice1 serving

Total for the whole meal

Energy: 535 kcal
Protein: 49 g
Carbs: 52 g
Fat: 14 g

Interactive illustration. Values are sample data, not a personalised nutrition target.

Check the pantry before shopping

The pantry records what is already at home; the shopping list records what still needs buying. Recipes sit alongside both in the same area of the app.

FROM RECIPE TO SHOPPING LIST

A little less to buy.

IngredientAt homeTo buy

Rice150 g needed

150 g—

Lemon½ lemon needed

½ lemon—

Fish200 g needed

—200 g

Broccoli120 g needed

—120 g

Rice and lemon are already in the pantry.

Illustration with sample quantities: check what is already at home before buying more.

Pantry stock also informs plan generation and budget estimates. A budget estimate considers missing ingredients after available stock, with coverage and limitations when prices or quantities are incomplete. It is an estimate, rather than a guaranteed checkout total.

Pantry with sample lemon, rice and olive oil stock grouped by category — The sample pantry already has rice, lemon and olive oil. Fish and broccoli are on the shopping list.

Shopping list with fish and broccoli to buy, using the same ingredient illustrations as the recipes — The sample pantry already has rice, lemon and olive oil. Fish and broccoli are on the shopping list.

Make waiting and starting feel considered

The quieter states are part of the experience too. While a new plan is being generated, a magnifying glass moves across illustrated ingredients. Progress and a short stage description explain the work happening behind it.

Loading an existing collection uses a different treatment. Shimmering placeholders follow the shape of the content that will appear, keeping the page recognisable while data arrives.

Play the recordings to see two kinds of waiting: composing a new plan and loading a recipe collection. Service timing is simulated.

Empty screens have their own illustrations and a clear next action. With no plans, the app offers a route to create one. An empty pantry has an action to add groceries.

The empty Plans screen with an illustration and an action to create the first plan — Empty states show the next useful step. These captures use a deliberately empty demonstration account.

The empty pantry with its own illustration and an action to add groceries — Empty states show the next useful step. These captures use a deliberately empty demonstration account.

Read saved plans offline

The mobile app is built with Expo and React Native. Supported data is stored locally in SQLite and refreshed from the server, so previously loaded plans and recipes can be read without waiting for a new request. Supported changes can be queued for synchronisation. Generating a new AI plan still needs a connection.