The pocket chef: how ChatGPT cooks dinner from "2 eggs and half a tomato"
Between calls, brain fried, fridge nearly empty — instead of ordering delivery: a chef prompt that asks clarifying questions and offers three options from simple to surprising.
01 — The situation
Sound familiar: a fridge exists, dinner doesn't?
Yesterday was rough. Call after call, a launch, a shoot, brain turned to mush. And between calls I'm standing at the fridge. Inventory: 2 eggs, half a tomato, an avocado, one lonely garlic clove, some ham.
I close the fridge and think: "forget it, I'll order pizza."
Then I remember the challenge I myself am running — and decide to test one simple thing. I just type the list of ingredients into ChatGPT with a prompt. And you won't believe it: it didn't just produce a recipe (a tasty one, by the way). It also asked: "do you happen to have spices? cheese? butter?" — and suggested how to make the dish better.
It looks like a small thing. But this is exactly where the "AI is complicated and for work" barrier breaks: no — it's also the thing that saves dinner in 30 seconds.
02 — The mechanics
Why is "what can I cook with eggs and a tomato" a weak request?
"What can I cook with eggs, a tomato and an avocado?" → The first omelet recipe off the internet. Ignoring what else you have, what you like and how much time you've got.
"Act like a real chef: first ask clarifying questions, then 2–3 options from simple to unusual, then step-by-step — and explain why the combination works" → A dialogue that actually makes you want to cook.
It's the same difference as in work tasks: a weak request gets the "most likely" answer, while a prompt with a role and an order of operations gets a personal one. The clarifying questions aren't a formality: one "do you have oil and spices?" changes the recipe entirely.
03 — The prompt
What does the full chef prompt look like?
Here it is — put your ingredient list at the top:
I have: [list everything — vegetables to sauces, if you remember].
Come up with what I can cook from these ingredients.
Act like a real chef:
- first ask me a couple of clarifying questions — maybe I have
something else that would boost the flavor,
- then offer 2–3 options from super-simple to unusual,
- then describe the cooking step by step, with comments on why
this combination works and how to enhance it
(with spices or texture, for example).And build an extended version around your habits:
04 — The task
Test it on your own fridge — right now
It genuinely takes 30 seconds — here's the whole task:
See what's there. List everything, including "one lonely garlic clove" and half a jar of sauce — the small things often decide the dish. Don't remember exactly? Approximate is fine.
Copy the prompt from section 03 (or build yours in the generator), paste the ingredient list, send. Answer the chef's clarifying questions — they're short: spices, oil, time.
From the 2–3 options, take the one matching your mood. Read the "why this combination works" comments — after a couple of weeks of such dinners you'll start assembling dishes without the prompt: taste literacy works in the kitchen too.
The prompt extends easily: "build a 3-day menu from these ingredients plus a shopping list," "adapt this recipe to what I have," "what can I cook for a kid who won't eat vegetables from this list." One mechanic: the chef role + clarifying questions + options.
05 — The point
What does this have to do with adopting AI?
Come on — a leftovers recipe, what does it have to do with "integrating AI into your life"?
Everything: the entry barrier breaks not on giant work tasks but on everyday wins. We think adopting ChatGPT requires three free hours, five tutorials and "unpacking the meaning." In reality that "later" never arrives — life always comes with deadlines and chaos.
Dinner from leftovers is the perfect first task: the cost of error is zero, the result is edible (literally), and your brain logs the experience "I set a task — the AI came through." Habits are built from such experiences — and then transfer to work.
Menu: "Build a 3-day menu from [ingredients] + a shopping list
with a rough budget"
Guests: "Guests arrive in an hour. From [ingredients] — what do
I serve to make it look festive? Ask clarifying questions"
Leftovers: "What do I do with [ingredient] that's about to go
bad? 3 options: eat today / prep / freeze"ChatGPT can come through in any situation — and dinner from "2 eggs and half a tomato" proves it faster than any tutorial. Take the chef prompt with its clarifying questions and test it on your fridge today. And share the strangest edible recipes you get.
FAQ
Are ChatGPT's recipes actually edible?
Yes — especially with the chef prompt: the clarifying questions filter out odd combinations, and the "why it works" comments let you judge a recipe before the stove. My 2-eggs-tomato-avocado dinner was genuinely good. Common sense still applies: a dubious step — ask again.
How is this better than "cook with what you have" recipe sites?
The dialogue: a site matches a recipe to 2–3 ingredients, while the chef prompt weighs everything at once — time, mood, restrictions, the lonely garlic clove — and asks clarifying questions. Plus on-the-fly adaptation: "no oven," "spicier," "for two" is just the next message.
Does it handle diets and allergies?
Yes — add restrictions right into the prompt (the generator has a field): "lactose-free," "gluten-free," "vegetarian." The model rebuilds the options. With serious allergies, do the final ingredient check yourself — that's a rule of any cooking, not just AI-assisted.
What everyday tasks should I try after the chef?
A multi-day menu with a shopping list, "what do I do with the ingredient about to go bad," a festive table from what's on hand. And then beyond the kitchen: the same "role + clarifying questions + options" mechanic works in travel, finances and planning.
