Hi everyone I used to work as a librarian in Korea — no coding background. I built Gwiraedang with Claude Code because I noticed ChatGPT and other AI chatbots often get Korean BaZi (사주) charts wrong: converting a birth date into the correct Four Pillars chart requires precise astronomical calculation (solar terms), and generative models tend to guess instead of compute. So I separated the two: a deterministic astronomy engine computes the chart (checked against Korea's official astronomical institute data), and 5 distinct AI counselor personas handle the conversation on top of an already-correct chart — each reading from a different classical tradition, designed not to just flatter you. Since I can't code-review myself, I leaned hard into automated testing (2,700+ tests) to catch mistakes instead. Would love feedback, especially on whether the "computed, not generated" pitch makes sense to people who've never tried a BaZi app before. Happy to answer anything about the build!

귀래당

Your Bazi chart, computed by astronomy — not guessed by AI

Your Bazi chart, computed by astronomy — not guessed by AI