Keep hitting Generate until you find a phrase you love, then:
Words are picked from curated lists using a cryptographically secure random generator. No patterns, no predictability.
An AI arranges those raw words into a proper sentence, adding prepositions, articles and a special character. Word soup becomes a vivid scene.
An AI generates a unique illustration of your passphrase. Your brain stores images far more effectively than strings of text.
Study the picture for ten seconds. Close it. When you need the password, recall the scene and the words come flooding back.
Cognitive scientists call it the picture superiority effect. Read a word and you have roughly a 10% chance of recalling it three days later. See an image and that jumps to 65%. By pairing a passphrase with a matching illustration you engage two memory systems at once: verbal and visual. This is dual coding theory, first proposed by Allan Paivio in 1971 and one of the most replicated findings in memory research.
"Frozen Merlin throws an anvil!" is almost impossible to forget once you have seen a cartoon of it happening. Your brain stores a scene, a story, a moment. The words are just labels for something you can already picture.
Password strength is measured in bits of entropy: how many guesses an attacker would need if they knew exactly how the password was generated. Each bit doubles the difficulty.
When the AI polishes a phrase it adds unpredictable structural decisions: which article? which preposition? where does the symbol land? singular or plural? These compound to add roughly 10 to 15 extra bits on top of the random word selection.
| Password | Entropy | Verdict |
|---|---|---|
| Password1 | ~10 bits | Cracked instantly |
| Fluffy2019! | ~18 bits | Seconds |
| Tr0ub4dor&3 | ~28 bits | Hours |
| The Crimson Owl? | ~27 bits | Hours (but memorable) |
| Merlin throws anvils! | ~42 bits | Weeks to months |
| Tiny Zeus juggles cacti! | ~53 bits | Years |
The key insight: a memorable phrase you can actually recall is infinitely more secure than a complex password scribbled on a sticky note.
Passy Pie is designed so that nobody, including us, ever knows your password.
store: false). Neither provider retains your prompt or the generated output for training or logging.secrets module), not a predictable pseudo-random source.In short: once you close the page, the password and image exist only in your memory (and your clipboard, if you copied them).
The words are derived from the EFF Large Diceware Wordlist, a carefully curated set of 7,776 common English words published by the Electronic Frontier Foundation. They have been sorted into categories (nouns, verbs, adjectives, colours, locations) and supplemented with historical figures, mythological characters, and public-domain fictional characters. No living people are included.
The word lists (wordlists.json), the AI prompt (prompt.txt), and the app configuration (config.json) all live in separate files so they can be expanded or tweaked at any time without touching code.