Why You Shouldn't Let AI Do Your Fortune Telling — And How to Do It Right

TL;DR — LLMs can't do calendar math. Code computes, AI interprets. That one line is the entire architecture. The first lesson I learned building a fortune-telling app: it's not about what you ask AI to do — it's about what you don't . This series is a build log toward launching a saju (Korean four-pillar fortune telling) app. Revenue is still $0. Real conversion and revenue data will come after launch. This isn't a success story — it's a debugging diary. "Tell me the fortune for someone born March 15, 1990." I threw this straight at an LLM. The response looked great. Smooth sentences, Five Elements this, Wood-Fire-Earth-Metal-Water that. But there was a problem. The base calculations were wrong. Me: "Analyze the Four Pillars for March 15, 1990, 6 AM" Claude: "The year pillar is Geng-Wu, month is Ji-Mao..." Me: "...Ji-Mao is wrong." LLMs can't do manseryeok (traditional Korean astronomical calendar) calculations. More precisely, they appear to get it right probabilistically, but they do