Skip to content
antara.
Explore my chart

Methodology

Keep the wonder. See the workings.

A calculation, an interpretation, and your own perspective. Here is how those layers fit together — and where their limits are.

On this page
  1. 01Planetary positions come from the ephemeris
  2. 02Chart facts are derived by classical rules
  3. 03Interpretations come from a curated knowledge base
  4. 04AI connects the claims into a reading
  5. 05Prediction accuracy has not been established
  6. 06Feedback records what was useful
  7. 07An honest note on what astrology is

  1. 01

    Planetary positions come from the ephemeris

    Every chart starts with the positions of the Sun, Moon, and planets computed from the NASA/JPL DE440 ephemeris. The calculation is separate from the language model. Astronomical position data does not establish the predictive validity of astrology.

  2. 02

    Chart facts are derived by classical rules

    Houses and planetary periods on your chart page follow deterministically from those positions by classical Vedic rules — whole-sign houses from the ascendant, Vimshottari dasha from the Moon’s nakshatra, and so on. Run the same birth details twice and you get exactly the same chart, down to the arcminute. With a saved birth-time range, you can also compare chart details at sampled times and see what changes. Unchanged samples do not prove stability between them; the reading still uses the entered time.

  3. 03

    Interpretations come from a curated knowledge base

    The interpretation layer uses a library of Parashari claims tied to chart configurations. The current seed library is internally drafted and is awaiting named expert review. Readings link to supporting claims using numbered citations, like this1. On a wide screen the numeral opens that source in a rail beside the reading; on a narrow one it opens the claim in place.

  4. 04

    AI connects the claims into a reading

    A language model receives your calculated chart and retrieved claims. It is instructed to use that material and acknowledge gaps. AI can still make mistakes: inspect the citations and their review status, and use the feedback controls when a reading misses the mark.

  5. 05

    Prediction accuracy has not been established

    Antara has no validated accuracy rate for personal predictions. A correct chart calculation, an expert-reviewed interpretation, and an event happening in someone’s life are three separate things. Scores and strength labels describe the traditional model; they do not measure the chance that an event will happen. Timing windows are heuristic estimates, and a narrower window does not make a prediction more reliable.

  6. 06

    Feedback records what was useful

    The thumbs under each reading record whether it was useful to you. They do not train the model, change evidence rankings, or establish predictive accuracy. Model improvements require reviewed examples and separate evaluation before release.

An honest note on what astrology is

Jyotish is a reflective tradition, and that is how we treat it. A reading is a structured lens for thinking about your life — not a prediction of it, and never a substitute for medical, financial, or legal advice. We make the calculations and cited claims inspectable so you can evaluate a reading for yourself.

Ready to see it in action? Find my chapter.