Questions

It is from दैवज्ञ, daiva-jña — daiva, fate or the divine, and jña, knower. The classical Sanskrit word for an astrologer, and literally "one who knows fate".

We picked it because of the second half. A daivajña was someone who had learned to do the calculation — the positions, the divisions, the periods — by hand, from tables. The claim in the word is knowledge, not prophecy. That is the distinction this whole product is built around.

No, and this is the part worth understanding. Your chart is computed by the Swiss Ephemeris — the same ephemeris professional astrology software uses — and stored. The transit view, the day forecast, the panchang and the PDF reports are all arithmetic over that chart. No language model touches any of them.

A model is involved in exactly one place: the chat, where it turns the computed chart into an answer in your language. It is handed a digest of your actual chart in its prompt, so it reads your planetary positions rather than recalling something vaguely astrological. When it needs something date-dependent — today's transits, a panchang, timing — it calls the engine instead of guessing.

Lahiri ayanamsa, whole-sign houses, and the mean lunar node for Rahu and Ketu — the standard North Indian defaults. Positions are topocentric, computed for your exact birth coordinates, which matters most for the Moon. The ephemeris is Moshier, which is accurate to well under an arcsecond for any date you are likely to ask about.

The Lagna moves about one degree every four minutes, and the Lagna sets every house in the chart. Four minutes of error can move a planet from your tenth house to your ninth, which changes the reading completely. Your Moon sign and nakshatra are far more forgiving.

If you do not know your birth time, say so when you create the profile. We record that rather than silently defaulting to noon and pretending.

Because a chart that cannot explain your past has no business predicting your future. We ask about a handful of real periods in the last few years, and check what the chart said was happening then against what you say actually happened. You get a percentage and a list of which areas of life verified and which did not.

It also does something more useful: if a birth time a few minutes either side of the one you gave fits your life better, readings quietly use the corrected chart. We never overwrite the time you told us — we correct the reading, not your memory.

A credit is one answered question, near enough. Billing is metered on the tokens a reply actually costs, but at the model tier we run, essentially every question lands at one credit. You are charged after the reply is generated, because the cost is not knowable before. A failed or empty generation is never charged, and a failure after a charge is refunded automatically.

The free tier grants 100 credits a month with no card. Beyond that: computing a chart is free, the transit view is free, the day forecast is free, and PDF reports are free, because none of them cost us anything to serve — they are pure computation. Only the chat spends credits.

No, and we tested that properly before deciding. Ten models answered an identical chart question, scored on tool-call survival, latency, cost and six objective markers true of that chart. A model costing a fraction of the most expensive one matched it on interpretation quality — because the precomputed chart already hands the model an authoritative answer. Paying more per token bought latency, not insight.

So the tiers differ on what you get — profiles, PDF reports — not on who answers you. Two models were disqualified outright for making zero tool calls, and one for confidently inventing a definition of a yoga that does not exist. That last one is exactly what this product is built to avoid.

Card payments are not switched on yet. In the meantime, promo codes work, referrals work, and we will add credits by hand if you email us. Your credits page has your referral link on it.

Your birthplace is looked up against a dataset on our own server, so no mapping service ever learns where or when you were born. Your questions and a digest of your chart do go to OpenRouter, which routes them to the model that answers — that is unavoidable for the chat, and it is the only part of the product where anything leaves our servers. Full detail on the privacy page →

For cultural, educational and reflective purposes only. Astrological readings are not scientifically validated and must not be used for medical, financial, legal or other consequential decisions.

We are precise about the astronomy because the arithmetic deserves to be right. That precision is not evidence that astrology predicts events, and we would rather say so plainly than let an exact-looking number imply otherwise.

The astrology engine and the Swiss Ephemeris under it are AGPL-3.0. Licences and attribution →