Budget apps usually respond to a receipt with categories, totals, and charts. Idiot Purchase imagines a less responsible but potentially more memorable response: read the purchase and roast the person who made it.
A receipt-scanning app concept that reads purchases and generates a humorous critique rather than another budget dashboard.
The aim. Explore whether mundane personal data can become a lightweight comedic interaction.
01The problem behind the project
The project tests personality, timing, and OCR as the value of a tiny app instead of feature breadth.
The project is a study in interaction voice. OCR and structured parsing make the joke possible, but the actual product is the timing and specificity of the generated critique—enough to feel personal without becoming cruel or exposing sensitive data.
People who enjoy self-directed purchase humor are the intended audience. Receipt subjects and merchants are affected by data handling.
02How it took shape
An iOS-oriented prototype combining receipt capture or entry with structured purchase parsing and AI-generated comedy.
The small iOS-oriented experiment combines receipt capture or entry, purchase extraction, and AI-generated comedy. Its public demonstration must use synthetic receipts because real examples can reveal payment identifiers, addresses, routines, and private purchasing history.
Josiah originated the interaction, tone, and product constraint and directed agent-assisted implementation.
A small prototype or project artifact exists, but its current source and reliability require review.
03What the project means now
The concept illustrates a recurring lesson in tiny apps: personality can be the feature. It also shows why novelty products still need strong privacy defaults and tone controls; a funny output does not justify retaining the document that produced it.
Receipts can expose payment, location, and purchasing details. Public demonstrations must use synthetic data, and generated insults need tone controls.
For a novelty app, a strong interaction voice can be more important than a large feature set.
Audit the surviving artifact and create a synthetic, privacy-safe demonstration if the joke still works.