24
로딩중...
Not an LLM prompt wrapper — a self-contained computation pipeline. From calendar calculations to ML ensemble, XAI interpretation, and MCP integration — every judgment is based on quantified data.
Understand yourself. Grow together.
Why am I this way — from your birth time alone, we read your energy, strengths, and the flow of your year.
What you miss alone becomes clear together — compare charts, ask @CAFE, and explore the depths of fortune-telling as a group.
Pro & API Partner programs available for practitioners and developers.
View Partner Program →Through a 6-step pipeline, every process is handled by our proprietary engine.
CAFE — an engine that quantifies Four Pillars astrology (Myungri), Korea's traditional natural philosophy (Cultural), through AI within an independent computational Framework.
Each model judges independently, then votes for the final decision. This prevents single-model bias and ensures stable predictions.
SHAP values decompose how key features contribute to the final judgment — stem combinations, branch clashes, element bias, seasonal fitness — providing quantitative verification. Six classical texts supply interpretive evidence, while 1,400+ real case studies cross-validate.
There are various schools of interpretation — focusing on structure, seasonal balance, and key elements. The engine computes major school perspectives in parallel and transparently discloses agreement or disagreement, so users receive balanced, unbiased analysis.
Six core classical texts are embedded in a Vector DB, automatically referencing original sources during analysis.
User queries are classified in 3 stages: Cache → Few-shot → LLM classifier for fast, accurate tool routing.
Based on Model Context Protocol, LLMs directly invoke 24Plus engine analysis tools. Structured data flows without intermediary translation.