Don't scan
this QR code.
You cannot read it. Your phone cannot read it either until it has already gone there. You were going to scan it anyway, and so is everyone else. That is true of every QR code ever printed. This one just admits it.
No account. About thirty seconds. Scan it with your own phone when you are done.
Twenty nine modules. It is this plain because the question lives here rather than inside the code, which is the trade the two modes are about. On a laptop, just click it.
What it actually is
A QR code that carries a question, not a link.
Everything else routes you to a page you then have to read. This one opens the AI you already use, with a good question about the thing in front of you already written out. You read it, and you press send.
You write the question once
Pick a recipe or write your own. Add the facts an AI would need to answer it well, since it cannot see the object.
We turn it into a code
Seal the whole question inside the code, or keep it here and put a few characters in the code. You download it, print it, stick it on the thing.
Someone scans it and taps once
Their own ChatGPT, Claude or Perplexity opens with the question already in it. They read it before they send it, so nothing goes out on their behalf.
The seal
Three layers, and each one is a claim we can prove.
Most QR products put a redirect behind a short link. Anyone who scans it can read the destination, copy it, and use it without you forever. A sealed code cannot be used without the key, and the key is here.
Sealed
On by defaultThe code carries ciphertext, not your prompt. Point any scanner app at it and you get a string of capitals. It is inert paper until it reaches us, which means nobody lifts your prompt off a photograph of your sticker.
Private
One toggleThe prompt is stored encrypted, under a key of its own that is itself wrapped by the server key. Only the wrapped form is ever written down, so a stolen copy of the database is a file full of noise.
Locked
For codes in publicA passphrase you hand out. The decryption key is derived from it rather than checked against it, so there is no lock to pick: without the key the prompt cannot be recovered by anyone, and that includes us.
The same idea, sealed. Sixty one modules instead of twenty nine, because this one is carrying the whole question.
Show me what is actually inside it
This question was encrypted inside the QR code you just scanned, and only qrprompter.com could open it. In under 80 words, explain why that is different from a QR code that holds a link.
Only this server could have told you that. Those words are not in the code in any readable form, and the code does nothing anywhere else.
What the seal does not do
When someone taps through, their assistant receives the prompt as text, because no assistant can decrypt anything. So a scanner who goes looking can read the prompt in that final address bar. The seal protects the printed code and forces every scan through us. It does not hide the prompt from the person holding the phone, and anyone who tells you otherwise is selling you something you can disprove in one scan.
Start from a recipe
The prompt is the product. These are the good ones.
A blank box is where most people give up. Each of these is a working prompt plus the reason it is written that way, because the difference between a code people scan twice and one they scan once is entirely in the wording.
Gym machine
I am standing at a seated cable row machine at the gym and I have not used one before. Give me the setup in three short steps, the…
Why it works. Names the exact machine, gives the model the situation it cannot see, and caps the length. A beginner at a machine has about twenty seconds of patience.
Wine bottle
I am holding a bottle of 2021 Argentinian Malbec from Mendoza. Suggest three dishes that go with it, one of which should be someth…
Why it works. The 'something I already have at home' clause is the whole recipe. It turns a generic pairing list into something the person can act on tonight.
Restaurant menu
I am looking at a menu with these dishes: [list your dishes here, with prices]. I want something [ask me what I am in the mood for…
Why it works. The only recipe here that makes the model ask first. A menu recommendation with no idea what the person wants is a coin flip, and the two questions cost four seconds.
Product packaging
I just opened a [product name] and I want to get it working without reading the manual. Walk me through first-time setup in five s…
Why it works. Replaces the leaflet nobody reads. The second question is where the value is: it is the thing your support team answers forty times a week.
Real estate sign
I am standing outside a house for sale at [address], listed at [price], [beds] bed and [baths] bath, built in [year]. Tell me what…
Why it works. A sign can hold six facts. The buyer's real question is 'is this a good price and what am I not seeing', and that is a question the agent cannot answer neutrally.
Plant tag
I just bought a fiddle leaf fig. Tell me exactly where to put it in a normal home, how often to water it, and the first three sign…
Why it works. 'Assume I have killed a plant before' sets the level in six words. Diagnostic signs beat a care schedule, because people do not look anything up until it looks sick.
Make one and scan it yourself.
That is the only demo that counts. It takes about thirty seconds and it is free.
Open the studio