ChatGPT Kundli — but with real maths
A general chatbot can discuss astrology fluently. What it cannot do is work out where the planets were when you were born. Here the chart is computed first, and only then read.
- Real ephemeris, not recall
- Sidereal & Lahiri
- Free chart, 3 free questions a day
Why generic chatbots get kundlis wrong
Ask a general-purpose chatbot where Saturn stood on the morning you were born and it will answer. The answer will be phrased with total confidence, in the right format, using the right vocabulary — and it will usually be wrong.
This is not a flaw anyone can prompt around. Working out a planet's position means evaluating a periodic series with hundreds of terms, correcting for light travel time and aberration, and then shifting from the tropical to the sidereal zodiac. It is arithmetic, not recall. A language model predicting the next token has no mechanism for doing it, so it produces something shaped like an answer.
The second problem is quieter. Most astrological text on the open web is Western, and Western astrology uses the tropical zodiac. Vedic astrology uses the sidereal one, and the two have drifted about 24 degrees apart. That gap is smaller than a sign, which is exactly what makes it dangerous — the answer lands one sign off and still reads as reasonable.
The pattern that does work
Stop asking the model to calculate. Get your placements from something that uses a real ephemeris, then hand them over and ask for interpretation only:
“My lagna is Cancer. Moon in Scorpio in the 5th, Saturn in Libra in the 4th, currently running Jupiter mahadasha. What does this suggest about a career change this year?”
That plays to what language models are genuinely good at — synthesis, explanation, holding several rules in view at once — while keeping them away from the part they cannot do.
It is also exactly how this site is built. Your chart is calculated here, in your browser, before any AI is involved. The model then receives those figures as fixed facts and is instructed never to compute a position itself. If something is missing from the chart, it has to say so rather than fill the gap.
A general chatbot vs a computed chart
The difference is not the quality of the writing. It is whether the numbers underneath are real.
- Planetary positionsRecalled from training data — plausible, frequently wrongComputed with VSOP87 and ELP planetary theory
- Zodiac usedUsually tropical, the Western defaultSidereal, Lahiri ayanamsa — the Vedic standard
- AscendantGuessed, if offered at allCalculated from your birth coordinates and sidereal time
- Dasha periodsApproximated, dates rarely reproducibleFull Vimshottari sequence from your Moon's nakshatra
- When it doesn't knowAnswers anywaySays so — the model may only use computed figures
To be clear about what is and is not being claimed: the calculations are verifiable astronomy. The interpretation on top of them is astrology, and remains a traditional practice rooted in belief.
ChatGPT and kundli — common questions
What a general chatbot can and cannot do with a birth chart.
Get the chart right first
Free, unlimited, no sign-up — and then three AI questions a day about what it actually says.