Conversational · Grounded
Ask your chart.
Enter your birth data, then ask anything. The model can only reason from the actual placements in your chart and must cite which ones informed each claim — no hallucinated signs, no invented degrees. Choose Western tropical or Vedic sidereal.
The chat opens after your chart is loaded.
Fill in the form above — birth date, time, city, and whichever tradition you read in. We’ll compute the placements; then the conversation begins.
Common questions
About asking your chart
How is this different from ChatGPT giving astrology advice?
Two critical differences. First, the chart itself is computed — sign, degree, nakshatra, house, aspects, dashas — using first-principles ephemeris math (VSOP87D + ELP2000), not guessed from a date. Second, the LLM is bound by a constraint: it may only reason from the placements we pass it, and must cite which placement informed each claim. It can't invent a sign, shift a degree, or hallucinate an aspect. If it can't answer from the facts, it has to say so.
Western or Vedic?
Both. Toggle the tradition on the form. Western (tropical) uses the zodiac aligned to the equinoxes — Aries begins at the spring equinox. Vedic (sidereal) uses the zodiac aligned to actual stars — about 24° behind tropical today, with Moon nakshatras layered on top. The underlying chart math differs; so does the interpretive lens. Pick whichever tradition you actually work with.
What does the assistant see about my chart?
After each answer, expand the '... placements the assistant reasoned from' panel to see the exact fact block that was sent to the model. These are the computed positions — sign, degree, nakshatra/pada for Vedic, house, aspects — and they're the only source of truth the LLM was allowed to use. If something you expected isn't there, it's a signal to ask a more specific question or to compute a fuller chart.
Why the 10-questions-per-hour limit?
Two reasons. First, Claude (the underlying LLM) has real token costs and we're running this as a free tool — the cap keeps it sustainable. Second, astrology rewards sitting with an answer more than stacking rapid follow-ups. If you're hitting the limit regularly, you're probably skimming; try reading the citations and the fact block, and the quality of the next question tends to improve on its own.
Is my chart data stored?
The engine is stateless — it does not persist your birth data or conversation history. Your conversation stays in your browser's local storage (keyed to your chart) so you can reload without re-asking. Clear it anytime using the Clear button above the conversation. If you enter a chart to ask a question, the server-side rate limiter records only your IP address and the time of the request, with no chart content attached.
How the grounding works
When you submit a question, the engine first computes a structured fact block from your chart: sign, degree, nakshatra with pada (for Vedic), house, ruler, aspects, current dasha. That fact block is sent to Claude along with your question. The model is instructed to answer onlyfrom those facts and to cite the specific placement behind each claim — you’ll see the citations highlighted in the answer, and the full fact block revealed below each response. If the facts don’t contain enough to answer, the model is instructed to say so plainly.