It's Sunday night. You're staring into the fridge, trying to answer a question that somehow feels both simple and impossible: What am I going to pack for lunch this week?

You've done turkey sandwiches three weeks in a row. Your kid is over it. You're over it. And yet here you are, reaching for the same deli pack because you're exhausted and out of ideas.

We built Lunchin because we lived that exact moment. And the answer wasn't just "more recipes"—it was a system that actually learns your family and gets better every week.

Here's how it works under the hood (without the tech jargon).

It Starts With Your Kid's Profile

When you set up Lunchin, you tell us about each child:

• •Allergens — Peanuts, dairy, gluten, whatever's on the list. These are hard blocks. The AI will never suggest something that contains them.
• •Likes — Strawberries, cheese, pasta. The foods they actually eat without a fight.
• •Dislikes — Mushrooms, anything "too squishy," spicy stuff. We steer clear.
• •Picky tags — This is where it gets specific. "No food touching." "Crunchy only." "Doesn't like sauces." These shape how meals get assembled, not just what goes in them.
• •Nutritional goals — High protein? Lower sugar? More veggies? We factor these in too.

This profile is the AI's starting memory. But here's the thing—preferences change. Kids are unpredictable. That's where feedback comes in.

The Secret Sauce: Your Daily Feedback

Every meal in your weekly plan has three simple buttons:

• 👍Liked it — Ate it up, want more of this
• 👎Disliked it — Came home untouched or with complaints
• ⭕Uneaten — Just didn't eat it (no drama, just data)

Here's what makes this different from a recipe app: we remember everything, but we weight it by time.

That thumbs-down you gave to "Veggie Quesadilla" two weeks ago? It matters. But the thumbs-down from six months ago on "Turkey Meatballs"? That matters less. Kids' tastes evolve. What was gross in September might be fine in February.

This is called time-decay learning—a well-established technique in recommendation systems [1]—and it means Lunchin adapts with your family instead of getting stuck on old data.

Picking Your Planning Personality

Not every family approaches lunch the same way. That's why we built four distinct "AI personalities" you can choose from:

You pick your personality in settings, and everything we plan gets filtered through that lens.

Behind the Scenes (Without the Jargon)

You might be wondering: what AI actually powers this?

We use a combination of leading AI providers—including Google, Anthropic, and others—with smart fallback routing. That means if one provider is slow or overloaded, we automatically switch to another. You just see fast, reliable meal plans.

The AI doesn't just spit out random recipes. It follows strict rules:

• ✓Never include allergens (ever)
• ✓Respect dislikes and picky tags
• ✓Use your feedback history to boost "proven winners"
• ✓Vary templates (bento boxes, wraps, thermos meals) to keep things interesting
• ✓Fit your prep time and budget preferences

And the prompts we use? They're detailed, opinionated, and constantly refined based on real family feedback. Each of those four personalities above has its own specific "brain" with different priorities.

What This Means for Monday Morning

All of this tech is invisible when you're using Lunchin. You don't see the time-decay algorithms or the AI routing. You just see:

• →A 5-day menu that makes sense for your kid
• →Items you can swap with one tap if something doesn't work
• →A grocery list that's already consolidated
• →A plan that gets better every single week

That's the goal. Less Sunday night panic. More Monday morning confidence.

Because you shouldn't need a degree in meal planning to feed your kids well.