Choosing what to eat with IBD can be complicated. Foods that feel safe one day might not work the next, and planning meals that match your disease type, symptoms and personal preferences can take more time and energy than you want to spend.

This is one of the reasons why IBDComfort uses AI to help generate recipes. But the process is not random, and it is not fully automated. A lot happens behind the scenes to make sure the meals you receive are simple, safe looking and relevant to your profile.

Here is a look at how it actually works.

AI creates the initial recipe collection

When we first started building IBDComfort, generating recipes took around 7 minutes per meal. That was far too slow to be useful in a real app. So instead of generating meals on the spot each time, we use AI to create a large pool of recipes in advance.

These recipes follow clear rules:

• simple ingredients
• gentle preparation methods
• no overly complex combinations
• balance between relapse friendly and remission friendly meals

This gives us a wide base of meal options without making the user wait.

Recipes are reviewed and stored in the database

After AI generates a set of meals, we check them and store only the ones that make sense. Some meals are removed because:

• the ingredient list does not match IBD friendly guidelines
• the preparation is too complicated
• the food pairing is questionable
• any other reason that we deem them not fit for you

Only the best and most consistent meals end up in the database.

It is important to say that we are not medical professionals, and this process is human driven. That means mistakes can happen. If you ever see a meal that does not feel right for your situation, we encourage you to report it through the Send Feedback section in the app.

As IBDComfort grows, our long term goal is to collaborate with gastroenterologists, medical experts and nutrition professionals so they can contribute their own meal ideas and help us expand the recipe base in a more guided way. This is not part of the app today, but it is a direction we care about.

Your profile filters the meals

This is where personalization begins.

When you create a profile, you set:

• your disease type (UC, Crohn or None)
• your current status (remission or relapse)
• foods you want to exclude
• foods you’re allergic to

These settings act like filters.

The app serves you only the meals that match your profile and needs.

Then the selection algorithm chooses meals for your daily plan

This part is not AI driven. It is a custom algorithm designed to:

• avoid repeating the same meals too often
• give you variety without overwhelming you
• select meals that match your current status
• respect your ingredient exclusions
• keep plans simple on flare days and more flexible on remission days

In other words: AI generates the meals, but the algorithm decides which ones you actually see.

This is what keeps the plans predictable and tailored to you.

Why we do not generate meals in real time

Real time generation sounds smart, but in practice it causes problems:

• it is slow
• the results are inconsistent
• meals might vary too much from day to day
• the user cannot rely on stability

By generating meals in advance and curating them, we keep the experience:

• fast
• stable
• consistent
• aligned with your profile

Your plan still feels personal, but without the randomness of raw AI.

How the database grows over time

Every now and then, we add new meals.

The process repeats:

• generate
• review
• store
• filter
• serve through the algorithm

This means the app gets smarter and richer without adding complexity for you.

In summary

AI helps create meals.
Humans review them.
Your profile filters them.
The algorithm selects them.
And you get a plan that is simple, predictable and built around your needs.

IBD is complicated, but choosing what to eat does not have to be.

If you want to see how this process looks in practice, you can try IBDComfort for free.