The thought never arrives in the kitchen. It arrives at the bus stop, in the lift at work, or forty minutes into a drive, and it is always the same shape: I think I turned it off, but I cannot picture doing it. Your phone can show you a photograph of the hob. What it cannot do, in almost every home, is answer the question without you staring at a dark room on a small screen, trying to decide whether that knob is at twelve o clock or at two.
Home automation is built out of sensors that measure one number cheaply. A contact sensor measures whether two magnets are near each other. A motion sensor measures whether infrared changed. A smart plug measures watts. Each is excellent at the thing it measures and blind to everything else, which is fine for a door and hopeless for a hob.
Consider what you are actually asking. Is anything on the stove right now. A gas hob draws no current worth measuring, so the plug approach is dead on arrival. A built in electric hob is wired to its own circuit, so there is no plug to insert. Even where a plug fits, under a countertop hotplate, it reports power and not meaning: a warming plate on low and a charger in the same extension lead are both a handful of watts. And the smoke alarm, the one device in the room designed for this failure, is deliberately late. It speaks when there is smoke, which is long after the pan went dry, and it speaks into an empty flat.
The gap in one line. Every device in your kitchen reports an event or a number. The thing you worry about is a state, visible in a photograph, that nothing in the room is looking at.
Stand in your kitchen doorway and look at the cooker. Without touching anything you can tell, in under a second, whether it is on. You are not sensing heat. You are reading four visual facts, and a vision model reads the same four from the same photograph.
| What you look at | Off looks like | On looks like |
|---|---|---|
| Knob position | Pointer at the neutral mark, all knobs aligned the same way | One knob rotated out of line with the others |
| Indicator lamp | Dark | Lit, usually orange or red |
| The ring itself | Grey ceramic, no flame | Blue or orange flame on gas, a glowing ring on ceramic or halogen |
| What sits on top | Bare hob, or a cold pan pushed to the back | Pan or kettle centred on a ring |
That is the whole trick. The app takes a still photograph on a schedule you choose, sends it to a vision model with the sentence you wrote, and hands you back an answer in plain words. No feed to watch, no clip to scrub through, no machine learning setup and no training photographs of your own cooker.
Scenarios are written the way you would explain the job to a neighbour looking after your flat. The more specific the description of what normal looks like in your kitchen, the fewer pointless alerts you get.
Each of those is a separate check with its own schedule, and they can share one phone. The second is the version families use for a relative who habitually leaves a pan out but never leaves a ring on, because it removes the alert that would otherwise fire every single evening.
Privacy, since it is a kitchen. The phone takes a still photograph on the schedule you set. There is no continuous recording, no stored video, and nobody is watching a feed of the room. What leaves the flat is a sentence about a hob.
Checks are paid for one at a time, which is why the schedule matters more than anything else on this page. The starter pack is 2000 credits for 2.99 euro, the light model costs half a credit per check and the fast model one credit. New accounts get 50 free credits without a card.
| Schedule | Checks per day | Credits per day (light) | Roughly |
|---|---|---|---|
| After you leave and at bedtime | 2 | 1 | about 50 cents a year |
| Hourly, 08:00 to 20:00 | 12 | 6 | about 30 cents a month |
| Every 30 minutes, round the clock | 48 | 24 | about 1.10 euro a month |
| Every 10 seconds | 8,640 | 4,320 | not a kitchen schedule |
The bottom row is on the table on purpose. A hob does not change between one second and the next, so paying for a check every ten seconds buys nothing but a bill. The two check schedule at the top covers the thought that actually follows you out of the door.
Can a camera tell me if I left the stove on? It reads the knobs, the indicator lamp, the flame or glow, and whether a pan sits on a live ring, and it answers in a sentence. It cannot sense heat, which only matters on an induction hob that shows no light at all.
Why is a smart plug not enough? Gas draws no current, a built in electric hob has no plug, and where a plug does fit it reports watts rather than intent.
Will my smoke alarm not catch this? The alarm is the last line and it is late by design. It speaks when there is already smoke, and it speaks into an empty flat.
How often should it check? Fifteen minutes after you normally leave, and once at bedtime. Hourly through the day if you are watching over someone who cooks alone.
Can it turn the stove off? No. It tells you, or fires a webhook, or speaks through the phone. Knowing in time is the whole product.
Does it work for an elderly parent? This is the case it fits best, because it asks nothing of the person being helped: no app to open, no wearable to charge, no live feed of their kitchen.
The kitchen check is rarely the only one. The two it is usually paired with are is the door actually locked, the same state versus event problem on a different object, and a quiet check on a parent living alone. If running cost is what you are weighing up, the cost per day arithmetic is worked out in full on its own page.
Stand a spare phone where it can see your hob and the knob row, type one sentence describing what off looks like in your kitchen, and set it to check twenty minutes after you normally leave. Most mornings you will never hear from it. The morning you do is the morning it paid for itself.
Get it on Google Play Download on the App Store Open in your browser