Shapes connect a grid representing a weekly meal plan with a grocery basket containing organised ingredients.
Ways to Buildthrough Where the Work Breaks

Building a Family Service That Connects Meal Decisions to Grocery Shopping

One case shows that even a non-developer can build a meal-planning prototype with AI coding tools.

Published 2026. 11. 8.

How a family frustration became a working screen

A case has emerged showing that even a non-developer can build a meal-planning prototype with AI coding tools. Torsten Kilgenstein in Germany demonstrated this by building a simple family meal tool his wife wanted.

He said he had no professional development background beyond some experience with web-page syntax around 20 years ago. He used Lovable, which creates software from written requirements, and ChatGPT, which helped him shape the plan, to create the screens, features, name, and introductory copy.

He disclosed monthly tool costs of US$20 for ChatGPT and US$50 for Lovable. Together, that is roughly KRW 100,000 a month.

He launched a working public beta of ForkCast. However, he did not disclose its price, user count, or revenue. The case shows that one person can launch a product, but not yet that people will keep using and paying for it.

The conditions for connecting meal planning and grocery purchasing are also growing in Korea. In a 2025 survey by the Korea Rural Economic Institute (한국농촌경제연구원), a government-funded agricultural and rural policy research institute, 16.3% of respondents said they mainly bought groceries online, up from 9.7% the previous year. Of KRW 37.8184 trillion in online food and grocery transactions recorded by Statistics Korea (통계청) in 2025, KRW 30.4039 trillion came through mobile phones.

But adding a purchase button does not connect the whole job. The Korea Consumer Agency (한국소비자원), a public consumer-protection body, reviewed 144 complaints about dawn-delivery services: 21.5% involved delivery delays, 18.1% product defects, 15.3% incorrect deliveries, and 10.4% missing items. Even when a menu is decided well, dinner preparation returns to manual work if key ingredients do not arrive.

Map where the work breaks first

Imagine Mr. Kim, a working parent of two elementary-school children, is responsible for weekday dinners. Before leaving work, he asks the family what they want in the family messenger, finds recipes through a search box, checks a photo of the refrigerator, then opens a grocery-shopping app.

He transfers recipe ingredients to the cart one by one. If his spouse replies, “The child does not eat mushrooms,” he searches for another recipe. If he cannot remember whether there is soy sauce at home, he either orders a duplicate or stops at a convenience store after getting home.

There are at least four handoffs here. Family preferences are read in a messenger, a menu is chosen through search, ingredients are moved into a note, and products are found again in a grocery app. When one menu changes, both the cart and the family notice must be updated.

The first service does not need to solve all of this. It can connect just four jobs: collecting the cooking time available today, letting the family choose from three meals, combining the ingredients for the chosen meal, and removing ingredients already at home.

In the changed flow, Mr. Kim taps 10, 20, or 30 minutes of available cooking time on the first screen. Family members mark only meals they dislike from three options. Ingredients for the decided meal are combined into one grocery list, and each item is marked as already at home, needs buying, or can be substituted.

At the start, the service does not need an automatic connection to a major supermarket cart. Testing with a copyable or shareable list keeps the build scope and cost low, while showing whether people actually use menu decisions and ingredient organisation together.

Some work still belongs to people. A child suddenly refusing to eat that day, judging the condition of fresh food, checking allergies, and making the final substitution for an out-of-stock item are all family decisions. A service is easier to trust when it gathers information scattered across several places rather than taking decisions away.

How products in other countries connected broken jobs

AnyList in the United States puts the shared grocery list at the centre. Couples or families can view the same list, import online recipes, place them on a meal calendar, and send only the needed ingredients to the list.

Core features are free, while its paid household service costs US$14.99 a year, or about KRW 20,000 using the conversion in the source. Rather than building a large food-delivery network, it charges a low annual fee for moving work from recipe to list and notifying the family again.

Cherrypick in the United Kingdom connected menu choice to sending ingredients to a preferred supermarket cart. Based on purchase history, it also tried to reduce duplicate purchases by estimating which ingredients were likely already at home.

A case study published by Google Cloud describes the service as having more than 500,000 users. It says 58% of users cooked with fresh ingredients more often and 82% said they ate more healthily. But these are self-reported responses from service users, so they need to be examined again against actual purchase records and food-waste volumes.

Cookpad Mart in Japan connected online ordering to the journey home. When users ordered fresh food, they could collect it from refrigerated lockers installed in apartment buildings, convenience stores, stations, and daycare centres, allowing them to finish grocery shopping while picking up a child.

In 2020, more than 10,000 households were covered by apartment-building installations, and 92.8% of 695 survey respondents were satisfied with collection inside their building. However, Cookpad later stated in a disclosure that, because of logistics and supply constraints, it had not created a model that combined repeatability, scalability, and profitability. It is an example of how moving fresh food in the real world is far more expensive and difficult than building screens.

Four things you can build in this gap

1. Family dinner agreement board

  • A service that shows available cooking time and candidate meals, then lets family members exclude only the meals they dislike.
  • It is for working parents who ask elementary-school children for menu preferences through a messenger every weekday.
  • You can describe its functions in words to create a small prototype. It is a good place to start because you can test whether menu-decision time falls without connecting purchasing.
  • The first screen should ask, “How many minutes can you cook today?” and show 10-minute, 20-minute, and 30-minute buttons.

2. Shared handover board for rotating grocery shopping

  • A service that lets several people share items, quantities, acceptable substitutions, and purchase completion when they take turns grocery shopping.
  • It is for households where parents and grandparents take turns caring for children and buying dinner ingredients.
  • Because online and offline purchases both remain part of the work, there is a need for a service that connects handovers between people before connecting to a specific store.
  • The first screen should list items to buy today with options for “Any product is fine,” “This product only,” and “Do not buy if unavailable.”

3. Pet-care shift board

  • A service that records food, walks, medication, and toilet condition, then passes remaining work to the next caregiver.
  • It is for households where spouses or siblings divide dog care around commuting schedules.
  • Like a meal service, this is recurring work that depends on messenger conversations and memory, so the same build approach can apply.
  • The first screen should show “Food,” “Walk,” and “Medication,” along with the person and time of the most recent completion.

4. Parents’ hospital-visit handover board

  • A service where children jointly organise appointment schedules, required items, transport responsibility, and questions to ask next.
  • It is for adult children living in different areas who take turns helping older parents attend outpatient appointments.
  • Information is scattered across calendar apps, family messengers, and paper prescriptions. Simply connecting the handover, without suggesting diagnoses or treatment, can still be useful.
  • The first screen should show the next appointment date and location, companion, documents to bring, and questions for the doctor.

One thing to check today

Ask one person who decided family dinners during the past week for 20 minutes, and have them show the exact process of choosing a menu and buying groceries on their phone. If they move between two or more tools, such as a messenger, search, notes, and grocery shopping, and manually transfer the same menu or ingredient three or more times, it may be worth building a service that connects one of those handoffs.

Why this matters where you are

The tools, grocery retailers, delivery reliability, and household routines in your market may differ from the examples here. You can still check whether families move the same meal or ingredient between messaging, search, notes, and shopping tools. If they do, test a small shared decision and list workflow before attempting retailer or delivery integration.

Sources

6 sources

Every fact in this article came from the pages below. Check them yourself.

Building a Family Service That Connects Meal Decisions to Grocery Shopping | Prometheon