EverydaypotStart

Photograph
what you have.

Three dishes you can cook tonight — with the steps, the timing, and what to swap for whatever is missing.

Put them in the pot. Anything on the counter, or the bag straight back from the market — salt, oil, pepper, onions, stock cubes and rice are assumed.

No typing. No dish name. No English needed. If you can point a camera at it, it can read it.

One pot.

The question in a kitchen at six in the evening is not what you feel like eating. It is what can be made from what is already here. That answer changes every single day, and it depends entirely on what came back in the bag.

So the pot starts empty, and you fill it — not with a search, with a photograph.

It knows ugwu from waterleaf.

A general model cannot. It will call bitterleaf scent leaf and it will call palm oil vegetable oil, confidently, every time. So this one is not asked an open question. It is handed roughly 120 Nigerian ingredients, each with a line on how to tell it apart by eye, and it must answer with a name from that list or admit it cannot.

Ugwu. Ugu. Pumpkin leaf. Three names, one ingredient, one row. It answers in the words used at the market.

Then it pours into three.

Not forty results with pictures. Three, ranked, that you can start cooking now — because a list is homework and a decision is dinner.

When it is not sure,
it says so.

Demonstration

It is 54% sure about this one. Which is it?

Below 75% sure, it does not guess. It asks — one question per photo, never five. Two seconds of your attention beats a silently wrong answer, and a silently wrong answer is the worst thing a cooking app can do to you.

Your correction is not thrown away either. Every tap is a labelled example against a real photo, so the thing that reads your kitchen gets better at reading Nigerian kitchens specifically.

It will never
invent a jollof.

The dishes are not written by a model. Forty of them sit in a table — real ingredients, real steps, real timings, each one validated by somebody who actually cooks it.

The model is allowed to do exactly three things: match what you have against those dishes, work out what can stand in for what is missing, and scale the quantities to the number of people eating. It never writes a method. You would spot it instantly, and you would be right to never come back.

Row 04 · recipes table

Efo riro

South-west · 35 min · serves 4 · one burner

Ingredients

  • ●ugwu or shoko1 large bunch
  • ●palm oil1/3 cup
  • ●tatashe3
  • ●onion1 medium
  • ●crayfish, ground2 tbsp
  • ○locust bean (iru)1 tbsp
  • ○stockfish1 piece
  • ○assorted meat300g

Method, as stored

  1. Bleach the palm oil gently, 2 minutes, no longer.
  2. Fry the blended tatashe and onion until the water has gone, 12 minutes.
  3. Stir in the ground crayfish and iru, 2 minutes.
  4. Add the washed ugwu last. Fold, cover, 4 minutes off the heat.

You set three things.
It sets the table.

How long you have, how many are eating, and whether you have gas or one burner. Nothing else. Move them and watch the three dishes change — this is the real filter, running here on the page.

How long have you got?
How many people?
What are you cooking on?
Cook this first

Efo riro

South-west · 35 min · single burner

8/9ingredients already in your photo

SwapNo stockfish in the photo. Use the crayfish you have and add it earlier, with the tatashe.

Number 2

Egusi soup with ugwu

South-east · 45 min · gas

7/9ingredients already in your photo

SwapNo ogbono. Grind the egusi finer and use a little more of it — you will get the body without it.

Number 3

Vegetable sauce and rice

Everywhere · 25 min · single burner

9/9ingredients already in your photo

Nothing missingEverything this needs is in the photo.

Demonstration · authored example ingredients and dishes, not a live result

Questions,
answered short.

Straight answers
What can I cook with the ingredients I already have?
Photograph them and Everydaypot tells you. It names what it sees, matches it against 40 Nigerian dishes, and gives you three you can cook tonight — with the steps, the timing, and what to use instead of anything missing.
Does it actually know Nigerian ingredients?
It is built for them specifically. It can name roughly 120 ingredients and nothing else — ugwu, waterleaf, bitterleaf, scent leaf, egusi, ogbono, iru, crayfish, palm oil — each with a note on how to tell it apart by eye. It answers in the words used at the market, so ugwu, ugu and pumpkin leaf all mean the same thing to it.
What happens if it names the wrong ingredient?
If it is less than 75% sure, it does not guess — it asks you one question and you tap the right answer, and the three dishes are worked out again with your correction in place. That correction is kept as a labelled example, so recognition gets better over time. If it is confidently wrong about something, trust your own eyes and not the app.
Do I have to type anything, or know what the dish is called?
No. The only thing you have to do is point a camera at your ingredients. You do not need to type, to spell an ingredient, or to know the name of the dish before you find it.
Will it work if I only have one burner?
Yes. You tell it whether you are on gas or a single burner and anything that needs equipment you do not have is filtered out before you see it. You also set how long you have and how many people are eating, and the quantities are scaled for you.
Do I need to photograph salt, oil and seasoning too?
No. Salt, oil, pepper, onions, stock cubes and rice are assumed to be in the kitchen already. Photograph the things that change the answer — the vegetables, the protein, whatever came back from the market.
What happens to the photo I upload?
It is read to identify your ingredients, then deleted after 90 days. Any correction you make is kept without the picture attached. There are no cookies and no analytics on this site, and you can ask for anything held about you to be deleted.

What is in your
kitchen right now?

Point the camera at the counter and find out. Salt, oil, pepper, onions, stock cubes and rice are assumed — you do not photograph your pantry.