Yanhuo Shijian Turns Meal Planning Into a Structured, Adjustable Workflow

Website: https://www.yanhuofood.com

Do you ever get this feeling?

Every day when mealtime comes, a few questions start looping in your head:

• What should I eat today?
• What ingredients do I still have at home?
• Have I ordered takeout too many times this week?
• I want to eat healthier, but I don't want to eat boiled chicken breast every day.
• When I finally decide to cook, I can't remember what I'm missing when I go grocery shopping.

It looks like just having a meal, but actually doing it involves a long chain of decisions: "what to eat, what to buy, how to cook, is it enough for everyone, will it be too complicated."

Honestly, most of the time it's not that we can't cook, it's that we don't want to think about it.

So I built a website called Yanhuo Shijian (Hearth & Mealtime).

It's not just an AI chat tool that recommends a few dishes, nor is it a content site that only displays beautiful recipes. I hope it's more like a dietary assistant that can really lend a hand: first help you figure out the menu, then connect the recipe, ingredients, and shopping into a smooth process.

Why build Yanhuo Shijian?

The initial idea was actually very simple.

I realized that there's no shortage of recipes online. A quick search on short-video platforms brings up all kinds of "10-minute quick meals," "one-week fat-loss meals," and "lazy one-pot rice." But having a lot of content doesn't make decisions easier.

You bookmark a bunch of recipes, but when evening comes, you still don't know what to eat;

You seriously plan a menu, but the next day you give up because you don't have time, can't find the ingredients, or your family doesn't like it;

You grab a big bag of stuff at the grocery store based on intuition, only to get home and realize you didn't buy what you needed and bought too much of what you didn't.

The real trouble isn't lacking a recipe for a single dish; it's lacking a tool that can string the entire process together.

So what Yanhuo Shijian wants to solve isn't just "recommending a dish," but this complete chain:

Decide what to eat → Generate a menu → Adjust dishes → View cooking instructions → Organize a shopping list → Record feedback → Take fewer detours next time.

The "Yanhuo" (fire and smoke) in the name refers to that warm, lively feeling of daily life; "Shijian" (mealtime) is both the time for eating and, hopefully, a small space for thoughtfully arranging each meal.

What can Yanhuo Shijian do now?

1. Generate a smart menu based on real-life conditions

On the menu page, you can fill in conditions like the number of people eating, the number of days to plan, how many meals per day, preferred and disliked flavors, dietary restrictions, health goals, budget, and cooking time.

The system generates a structured menu based on this information, rather than just throwing a big block of text at you that you'll forget after reading.

For example, you can make requests like this:

• Two people eating, plan for the next 5 days;
• Weekday dinners should ideally be done within 30 minutes;
• Less oil, lighter flavors, but don't make every meal a salad;
• No cilantro, and one person can't eat very spicy food;
• Ingredients should be as common as possible, available at a regular market or supermarket.

If you're not satisfied with one dish after generation, you don't need to scrap the entire menu and start over. You can swap just that one dish, or replace the whole meal, and you can state reasons like "don't like it," "too complicated," "too expensive," or "had it recently."

This is very important to me, because real life rarely goes exactly according to plan. A truly useful menu shouldn't be unchangeable after generation; it should allow you to adjust at any time.

2. If you don't want to plan a whole week, just solve "today's meal" first

Sometimes people don't want to make a long-term plan; they just want to quickly solve the meal right in front of them.

"What to Eat Today" is prepared for this scenario.

You can tell it:

• What ingredients you have at home right now;
• How much time you have to cook today;
• Whether you want something light or something more satisfying;
• Whether you're tired today and how much effort you want to put in.

The system will give several options in different directions, such as one that better matches your current needs, one that's more hassle-free, and one for a change of taste. After selecting one, you can directly add it to the menu and continue to view the recipe or generate a shopping list.

Its goal isn't to make a single meal overly complicated, but to help you spend a few less minutes agonizing.

3. If you really don't want to cook, see what's nearby to eat

I later realized that dietary planning can't just consider "cooking at home."

Working overtime, business trips, last-minute meetings, an empty fridge, or simply not wanting to turn on the stove today—these are all part of real life.

So the site also has a "What's Nearby to Eat" page.

You can use your current location, or enter a company, campus, or other address, to search for real nearby dining spots, then filter by distance, budget, rating, and business hours.

If you still can't choose, you can let the system make a decision for you from the suitable results.

Of course, map information and business hours might have delays, and delivery range, real-time prices, and discounts should still be confirmed on the ordering platform. This feature isn't meant to replace map or food delivery apps, but to reduce the process of "opening the app, scrolling for ages, and still not knowing what to order."

4. After the menu is generated, automatically organize an ingredient list

Deciding what to eat is only the first step; what really gives people a headache is often the grocery shopping.

Yanhuo Shijian can extract the required ingredients from the generated menu and organize them into a shopping list by category.

You can:

• Select the menu and date range to shop for;
• Mark which ingredients you already have at home;
• Check off items you've already bought;
• View by category like vegetables, meat, seasonings, etc.;
• Copy, share, export to CSV or image, or print directly.

This way, when you go to the supermarket, you don't need to flip through the menu while trying to remember, and it can also reduce duplicate purchases and food waste.

5. Gradually accumulate frequently made dishes into a recipe library

Dishes in the menu can be viewed with more detailed ingredients, seasonings, cooking steps, time required, difficulty, and notes.

If you find a dish suitable, you can also add it to your own recipe library.

You can also record "liked," "made it," or "don't want to eat again." This feedback isn't just a mark; it can also be used as a reference when generating menus later.

I hope that the longer you use it, the more it becomes your own dietary tool, rather than starting from scratch to get to know you every time you open it.

6. The homepage isn't a promotional page, but a real workbench

The homepage of Yanhuo Shijian doesn't pile up a lot of flashy promotional content; instead, it tries to put common operations front and center.

If you already have a menu, you can quickly see what to eat today, continue adjusting the current menu, or open the shopping list; if you don't have a menu yet, you can start generating directly from common scenarios.

Previously generated menus are also kept, making it convenient to continue using or referencing them, without needing to re-plan every time.

I don't want to make it an AI product that "looks impressive but is unusable"

There are many AI-powered products now, and generating a large block of content from a single sentence isn't hard.

What's really hard is: can the user directly use what's generated for the next step?

So when building Yanhuo Shijian, I paid attention to a few things:

1. Results must be structured. A menu is a menu, ingredients are ingredients, steps are steps; they can't all be mixed into one block of text.
2. Partial modification must be possible if unsatisfied. Users shouldn't have to regenerate the entire week's plan just because they don't like one dish.
3. Real life must be considered. Time, budget, number of people, dietary restrictions, whether ingredients are common—these are more important than "looking fancy."
4. It must be executable. From menu to recipe, to shopping list, the process should connect smoothly.
5. Be honest about AI. AI can make mistakes, and nutritional data can only serve as a daily estimate, not a replacement for a doctor or professional nutritionist.
6. Constantly adjusting. Actually, it has already gone through several versions, mainly because I've been looking for a more suitable approach. If anyone has suitable suggestions, please feel free to raise them.

Simply put, I don't want it to only make people think "wow, so smart" the first time they open it. I hope that after some time, you're still willing to come back and use it to plan your next meal.

About privacy and model configuration

Yanhuo Shijian supports the public capabilities provided by the site, and also supports configuring your own model or map service Key in the current browser.

Configurations filled in by users themselves are only saved in the current browser and will not be written to the Supabase database. When calling services, they are only used for that single request, and the server side will not store these Keys long-term.

I know many people are sensitive about "whether my information will be stored," so data that doesn't need to be uploaded is kept locally as much as possible; where external services need to be used, the boundaries are explained as clearly as possible.

Since the project is still a demo validation project, there is basically no cost investment for now, and it is temporarily a public welfare project. Everyone is free to use it. You can either follow my WeChat public account to use my configured model and Amap Key for free, or you can configure your own, which is also only saved locally in the browser and will not be uploaded or exposed.

Who is it suitable for?

I think Yanhuo Shijian is more suitable for these scenarios:

• Office workers who agonize over what to eat every day;
• People who want to cook for themselves but don't want to spend too much time planning;
• Couples or families who need to accommodate different tastes;
• People with daily dietary goals like dietary restrictions, oil control, sugar control, or fat loss;
• People who often buy too much or miss items and want to organize their shopping list clearly;
• People who want to reduce takeout but don't want cooking to become a burden.
• Suitable for customizing some team meals.

Of course, if you only occasionally need to look up a recipe for one dish, a regular recipe website might be enough. Yanhuo Shijian is more suitable for those who want to arrange "what to eat, how to buy, and how to cook" all together.

It's not perfect yet, but I will continue to improve it

Yanhuo Shijian is still under continuous iteration.

AI-generated menus won't meet expectations 100% of the time, map data may not be completely real-time, and the nutrition and portion sizes of some dishes still need further calibration.

But these problems won't just be dismissed with a simple "AI is for reference only." Going forward, I will continue to optimize the stability of menu generation, recipe credibility, nutrition estimation, personalized feedback, and the mobile user experience.

If you encounter unreasonable menus, difficult operations, or have a feature you particularly want added while using it, you are welcome to raise it.

Finally

The original intention of building Yanhuo Shijian wasn't actually to have AI teach everyone how to live their lives.

I just feel that eating is a very ordinary, very concrete thing, but something we encounter every day. If a tool can help us agonize a little less, waste a little less food, and eat a few more suitable meals, then it already has value.

If you've also been frequently not knowing what to eat lately, feel free to come and try it:

Yanhuo Shijian: https://www.yanhuofood.com

GitHub: https://github.com/you-want/yanhuofood

You don't need to plan everything perfectly from the start. Just solving today's meal first is good enough.