An AI reading of the relationship between two people, combining their MBTI types with their birth dates in the Korean saju tradition. Entertainment, not advice.
This reads two people at once: each person’s MBTI type alongside the four-pillar 사주 chart derived from their birth date, with a language model writing about how the pair fits together. It is entertainment. MBTI is a self-report questionnaire without predictive validity in the psychometric sense, and 사주 is a traditional interpretive system rather than a measurement of anything — neither predicts how a relationship turns out.
What it is genuinely good for is a reading you can argue with. The text is generated per request rather than looked up in a compatibility table, so it responds to the specific pair rather than to the sixteen-by-sixteen grid, and the follow-up box keeps the context so you can ask one more thing without re-entering both people. Birth dates are read as solar dates, and nothing you type is stored here — it goes to the model that writes the reply and nowhere else.
The two systems make different kinds of claim, and it is worth knowing which is which. MBTI sorts answers into sixteen types along four axes; its own publishers describe it as a preference indicator rather than a measure of ability, and psychometric reviews have repeatedly found that a substantial share of people land in a different type on retaking it a few weeks later. 사주 is a traditional interpretive framework built on the sexagenary cycle, read from the year, month, day and hour of birth. Neither predicts how a relationship will go, and a reading that claimed otherwise would be making something up.
What both are genuinely useful for is vocabulary. Having a shared word for a preference makes a conversation about it possible: telling someone that you need quiet after a long meeting is easier when there is a name for it than when it sounds like an excuse. The reading here is worth treating the same way — as a set of prompts about how two people might differ, which you then check against the people rather than the chart. The parts that are wrong are often more useful than the parts that are right, because disagreeing with a description is a fast way to find out what you actually think.
On the mechanics: birth dates are read as solar dates, which is the distinction that matters most for 사주 accuracy, since a lunar birthday converts to a different day and therefore a different pillar. The text is generated for each request rather than looked up in a compatibility table, so it responds to the specific combination in front of it instead of a sixteen-by-sixteen grid, and the follow-up box keeps the earlier context so a second question does not require re-entering both people. Nothing typed here is stored by this site — it goes to the model that writes the reply and nowhere else.
The hour pillar is the part most readings quietly get wrong, because most people do not know the time they were born. 사주 divides the day into twelve two-hour branches, so an hour that is off by one puts the reading on a different pillar entirely — and a birth time recorded on a family register is often the time of registration rather than of birth. Where the hour is genuinely unknown, a reading built on the other three pillars is the honest version, and treating a guessed hour as fact adds detail without adding accuracy.
Not very. It is generated for entertainment and reflection, and neither MBTI nor saju predicts relationship outcomes. Enjoy it as a prompt for a conversation.
The page does not keep it. The details are sent to the AI model to produce the reading and are not saved on this site.
The underlying model is generative, so wording and emphasis vary between runs even for identical input.