How your reading is made

Your palaces, star placements, brightness, four transformations, pillars, and luck cycles are calculated by taibu-core 4.0.0 from your birth data. The engine returns a fixed, reproducible result for a given input and rule set.

Interpretive text is built from those computed facts. Sentences that state the facts are always written by code. When “Allow AI-written narrative” below is on, the reflective wording between those facts may be written by a language model (Google Gemini), which receives only the computed facts — never your birth data, name, or location. When it is off, or the model is unavailable, the reading falls back to a deterministic template: same chart, same sentences.

The model cannot add placements, dates, or predictions: it has no field for them, and every sentence it returns is checked before display. Each reading shows its Basis line, and ⓘ opens the full provenance trail including which parts were model-written.

When this is on, we send only computed chart facts to the language model — palace names, star names and brightness, transformation labels, pillar characters, cycle years, card names. We do not send your name, email, precise birthplace, or raw birth data. We never use your readings to train or fine-tune models; how the provider handles the request is governed by its API terms.

Turning this off costs you nothing: the free tier always uses the template, and Pro falls back to it.

If a reading is offensive, gives medical, financial, or legal direction, or states a fixed prediction, flag it. Reports reach a human queue and are reviewed within 24 hours. Content that breaks our guidelines is removed.

Kairoscope is for entertainment and self-reflection. It does not predict events, and it is not a substitute for medical, psychological, financial, or legal advice. If a reading troubles you, speak to a qualified professional.