Built a Vocabulary Mini-Program in 2 Hours with AI: Why Clear Rules Matter More Than Code

The author launched a free WeChat mini-program called "单词来咯" (Word Comes) for vocabulary learning. It has no in-app purchases. The development process involved writing less code than expected; most time was spent clarifying product rules so that WorkBuddy could implement them correctly.

Review Mechanism

The mini-program uses spaced repetition based on the Ebbinghaus forgetting curve. A word is automatically scheduled for review on days 1, 2, 4, 7, 15, and 30 — six time points managed by the system.

Four Question Types

New words are introduced via flashcards to associate form and meaning. Review words rotate among four question types (customizable): flashcard, choose meaning given word, choose word given meaning, and spelling.

Spelling was added deliberately because multiple-choice questions can create an illusion of mastery — recognizing a word is not the same as producing it. The choose-word and spelling types test the ability to recall and write the English word from the Chinese meaning.

Using WorkBuddy: The Real Time-Saver Isn't Typing

Writing UI, data storage, and question-type switching would take a skilled frontend developer a long time, but those tasks are deterministic — if you know what you want, you can code it.

The real effort lies in stating requirements without ambiguity. Human developers fill in common sense automatically; the tool executes literally. The key to effective collaboration is not writing clever prompts but thinking through product rules first. Any unclear rule becomes a bug in the implementation.

The biggest takeaway is not how many hours were saved, but the realization that writing unambiguous rules is harder than writing code.