Several meal planning tools will happily generate a full week of dinners you have never seen, built from a huge internal recipe database rather than anything you already cook. Kyura's Generate my week feature works the opposite way: it only ever selects from recipes already saved in your library. This is a deliberate choice, not a limitation we're apologising for, and it's worth explaining why.
The problem with an app that invents dinner
A generic AI-generated meal plan is disconnected from what's actually in your kitchen, what your family will eat, and the recipes you've already found and liked enough to save. It might suggest a dish you've never cooked, using an ingredient you don't have, cooked in a style nobody in your house prefers. Following it means either abandoning your existing recipe collection or running two separate systems: the AI's suggestions and the recipes you actually trust.
There's also a quieter cost. If you've spent time capturing recipe screenshots, saving cookbook photos, and building a personal collection, a generator that ignores all of that and invents something new every week makes that collection pointless. The library becomes a place recipes go to be forgotten rather than a resource the app actually uses.
What "Generate my week" actually does
The generator scores the recipes already in your library against the preferences you set for that week: dietary pattern, effort level, household size, ingredients to avoid, an optional calorie range, recent cooking history, and your grocery budget where that's set up. It then builds a draft week from that scored list. Weekend slots can favour slower, more involved recipes, while weekday slots can be weighted toward quicker ones. Nothing in that process pulls in a recipe from outside what you've saved.
The matching logic in the current version is rule-based rather than an opaque machine-learning model, which means the reasons a particular recipe was chosen for a particular slot can generally be explained rather than shrugged at.
The trade-off we accepted
The obvious cost of this approach is that a smaller recipe library produces a less varied generated week. If you've only saved eight dinner recipes, "Generate my week" cannot manufacture an interesting seven-day plan from nothing, and it won't try to. When your library can't fill every slot with something that matches your preferences, Kyura shows the gap directly rather than quietly loosening the rules or fabricating a recipe to avoid an awkward blank space. We think an honest gap is more useful than a plausible-looking week built from a recipe you've never actually cooked and might not enjoy.
Why this fits how people already cook
Most households, even ones that think of themselves as adventurous cooks, rotate through a smaller set of go-to meals than they'd guess. A tool that helps you use that existing rotation more deliberately, varying it, avoiding repeats within the same week, matching it to a budget, tends to be more useful day to day than one that keeps suggesting something entirely new.
Staying in control of the week it builds
A generated week is a draft, not a final answer. Any single meal can be swapped for another saved recipe, one day can be regenerated on its own, or the whole week can be rebuilt from scratch if none of it feels right. That review step matters as much as the generation itself: the goal is a starting point you can adjust quickly, not an instruction you're expected to follow exactly.
A note on cost awareness
Because the generator only draws from recipes with data you control, budget awareness follows naturally. If you've entered ingredient prices for the budget feature, the same weekly grocery figure can be taken into account when the generator scores recipes, favouring less expensive weeks without asking you to manually cost-check every option yourself.
What this means for recipe discovery
It's worth being direct about the trade-off one more time: if you're hoping a meal planning app will introduce you to entirely new dishes, Generate my week is not that tool, and it isn't trying to be. Discovering new recipes is still something you do deliberately, through screenshot capture, a cookbook, or a recommendation from someone else. The generator's job starts after that, arranging what you've already decided is worth cooking into a week that fits your preferences and your budget.
When the library isn't big enough yet
If Generate my week keeps showing gaps, the fix isn't a better algorithm, it's a bigger library. Capturing a few more recipes for the meal types or dietary needs where gaps keep appearing gives the generator more to work with next time. It's slower than an app that invents something on the spot, but the payoff is a week built from meals you've actually chosen to keep.