How CelestKin Computes Readings

1. Chart computation

The chart math behind every reading is produced by a server-side engine that wraps Swiss Ephemeris — the same planetary ephemeris used by most professional astrology software. From a birth date, birth time, and a geocoded latitude/longitude, the engine derives planetary positions, houses, aspects, Nakshatra boundaries, Dasha start dates, and the tradition-specific quantities each system needs. The output is deterministic: the same inputs produce the same numbers every time.

The engine covers nine traditions: Vedic (sidereal, Lahiri ayanamsa), Chinese (Four Pillars and lunar-new-year corrected animal signs), Mayan (Dreamspell Tzolkin), Pythagorean numerology, Western tropical astrology, KP sub-lord theory, Human Design (derived from astronomical gate positions), biorhythm, and a palm-reading framework that runs on uploaded images. Each system ships its own reference guides — the canonical texts our prompts draw from.

2. AI-assisted interpretation

Once the chart is computed, the numerical result plus the relevant reference guides are assembled into a structured prompt and sent to Claude (Anthropic) via AWS Bedrock. The AI is instructed to write in the voice of a reflective astrologer, cite the tradition it is drawing on, and avoid deterministic claims about outcomes. We use extended thinking for deeper analyses. Full details are in the AI Disclosure.

The AI sees your chart data for the duration of a single request. It does not retain anything between sessions, and AWS Bedrock does not use prompts or responses to train foundation models.

3. Editorial rules (the ten language laws)

Every prompt is constrained by a short list of editorial rules we call the ten language laws. They require the AI to frame results as reflection rather than fate, to name the uncertainty in its own output, to avoid prescriptive claims about money, marriage, medical, or legal matters, and to protect the reader's agency. When a reading sounds prescriptive or deterministic, we treat it as a bug and tune the prompt layer.

4. What CelestKin does not do

• Predict outcomes. No reading tells you what will happen. A transit or Dasha describes a theme the classical texts associate with a window — not an event.
• Diagnose. We do not produce medical, psychological, legal, or financial diagnoses. We are not a substitute for a clinician, a lawyer, or a financial advisor.
• Make decisions for you. Readings surface patterns and prompts. The decision stays with you.
• Claim validation. Astrology has no proven predictive power in the scientific sense. We use it as a structured journaling and reflection surface, and we say so.

5. Reflection-first design

The whole stack is tuned around a single question: does this reading help the user reflect, or does it push them to outsource? Deterministic language, absolutist claims, and prescriptive verdicts all fail that test. If you ever catch us breaking the rule, write in via the contact page.