--- title: "DashaGPT Vedic Astrology Chatbot: The Rules It Runs On" slug: "dashagpt-vedic-astrology-chatbot" canonical_url: "https://mahadasha.com/dashagpt-vedic-astrology-chatbot" description: "The DashaGPT Vedic astrology chatbot answers from your own chart and a retrieval index, never from recall. What it refuses to say, and what enforces that." published_date: "2026-08-06 00:00:00" modified_date: "2026-08-06 00:00:00" author: "Pawan Acharya" keywords: "dashagpt vedic astrology chatbot, astrology chatbot, ai astrologer chat, vedic astrology ai assistant, free astrology chat, ask an astrologer online" locale: "en" --- # DashaGPT Vedic Astrology Chatbot: The Rules It Runs On > A chatbot that answers astrology questions is only as good as what it refuses to say. Here is the refusal list, and the chart block it reads before every reply. # DashaGPT Vedic Astrology Chatbot: What It Is Allowed to Say The DashaGPT Vedic astrology chatbot is a conversation with your own chart attached. Before it writes a word it has already been handed your planetary positions, the dasha periods running today, and up to five passages pulled from a written corpus. Everything it says has to come out of that material, and the interesting part of the design is the list of things it may not do with it. This post is that list, along with what actually happens between pressing send and seeing a reply. ## What the DashaGPT Vedic astrology chatbot reads first Three blocks go into every request. The first is chart context: signs, houses, nakshatras, the dasha ladder, and whichever divisional charts the question touches. If you have no saved birth details this block is empty, and the assistant is required to notice and ask rather than answering in general terms and hoping you do not spot the difference. The second is retrieval. Your question is embedded and matched against a vector index built from the interpretation corpus and the site's own reference pages, returning 5 passages with no more than 2 from any single topic. That cap exists because without it a question about marriage comes back with five paragraphs about the seventh house and nothing about the period running over it. The index behind it holds 13,379 interpretation files across three languages, so the limiting factor is almost never whether something relevant exists. The third is memory. Facts you have mentioned in earlier conversations are stored and fed back, up to 50 per request. How long they survive depends on your tier: 30 days on the free one, 365 on the entry paid tier, and indefinitely above that. This is why it can remember that you changed jobs in March without being told again, and also why it will eventually forget if you are not paying for it. The three blocks are assembled fresh for every message, which is worth knowing because it explains a behaviour people find odd. Correct your birth details and the next reply is already using them. There is no cached reading of you sitting anywhere, and nothing to invalidate. ## The refusal list Four rules do most of the work, and they are worth knowing because they explain the moments when the assistant seems unhelpful. It may not calculate. Planetary positions, house numbers and dasha dates come from the chart block or they do not appear at all. A language model asked to work out where Saturn stood in 1987 will produce a confident number and it will be wrong often enough to matter. So it is never asked. It may not predict absolutely. Phrases like you will marry, you are guaranteed, or this will definitely happen are caught by a validator that runs over every response before it is stored, and rewritten into conditional forms. The rule is not about politeness. A period is a tendency, and the [measured base rates](https://mahadasha.com/event-timing-in-vedic-astrology) for the standard timing combinations are wide enough that certainty would be a false claim in any tone. It may not cite sources aloud. Retrieved passages are woven into the answer rather than labelled as source one and source two. That is a readability choice rather than concealment, since the passages are interpretive text and not references a reader could go and check. It may not sell to you unprompted. Reports, prices and offers come up only when you ask about them. An astrology answer that closes with a discount code is a product failure, not a feature. ## What the validator actually catches Every reply passes a checker before it is persisted. The checker looks for planets and signs that appear nowhere in your chart, for dasha dates that were not in the context block, and for the absolute-language patterns above. Matches are flagged, and in the case of the language patterns, softened. Its reach is worth being honest about. The validator can tell that a reply mentions Scorpio when no planet of yours is in Scorpio. It cannot tell whether the interpretation of a genuine Scorpio placement is any good. It catches fabrication, not weakness, and most of what makes a reading useful or useless sits in the second category where no automated check reaches. ## It also answers questions about the site A quirk of building an assistant inside a product is that people ask it product questions. Where is my report, how do I change my birth details, which payment methods work in my country. So the retrieval layer runs an intent check first and swaps the corpus: an astrology question searches the interpretation index, a support question searches documentation instead. Mentioning it here matters because the two modes fail differently. An astrology answer can be vague and still be reasonable. A support answer that is vague is simply wrong, and the fallback for anything it cannot resolve is to hand you to a human rather than improvise a policy. ## What it is genuinely good at Three kinds of question get noticeably better answers than the rest. Questions about your own chart's structure, because the facts are sitting in front of it. Which house does my tenth lord occupy, what is my birth star, which antardasha am I in this month. Questions that join two things you would otherwise look up separately. What is moving over my seventh house at the moment, and how does that sit with the period I am running. Retrieval is decent at pulling both halves and the chart block supplies the join. Questions where you want reasoning rather than a verdict. Ask why instead of whether and the answer improves, because the chart gives it something to reason from and the corpus gives it a tradition to reason within. What it is poor at is the thing everyone tries first, which is a date. Ask when something will happen and you get a period and a set of conditions. That is the true answer and it is almost never the satisfying one. There is a fourth category worth knowing about, which is questions with a premise built into them. Ask why your seventh house is weak and the assistant will answer inside that framing rather than checking whether it is weak at all. The chart block is in front of it, so the fix is easy: ask what the seventh house looks like before asking why it looks that way. Leading questions get led answers, and that is a property of the format rather than of this particular assistant. ## Guests, credits and the daily ceiling Without an account you get three messages a day, which is enough to find out whether you want one. Registering raises that to 10 daily and adds a wallet of 100 credits that does not expire, so a first month is unusually generous by design. Ordinary questions cost a credit each and a deeper analysis costs two. Above the free tier the daily ceiling gives way to a monthly budget, and the memory and timing limits described earlier both open up. The [pricing page](https://mahadasha.com/pricing) carries the current figures and this post deliberately does not, because a number written into an article is a number that will be wrong eventually. ## Trying it against something you can check The most useful first test is a question whose answer you already know. Ask which nakshatra your Moon occupies and compare it against the [Birth Chart Calculator](https://mahadasha.com/birth-chart-calculator) , or ask which mahadasha was running during a year you remember clearly. Agreement tells you the chart block is wired to your real data. Disagreement usually means the birth details on file are not the ones you think they are, which is a far more common problem than it sounds. Then open [DashaGPT](https://mahadasha.com/dashagpt) and ask something you actually want to know. ## Frequently asked questions about the assistant ### Is the DashaGPT Vedic astrology chatbot free? Three messages a day with no account, ten a day with a free one, plus a hundred-credit welcome grant that carries over. Paid tiers move you to a monthly allowance and a longer memory. ### Does it see my birth chart? Yes, when you have saved birth details. They are computed by the same engine the calculators use and injected as a context block. Without them it asks rather than guesses. ### Can it tell me when I will get married? It will describe the periods and transits that traditionally bear on marriage and refuse to name a date. A validator enforces the refusal, because the classical rules for marriage are satisfied across roughly two fifths of an average lifetime. ### What language does it answer in? Whichever one you write in. The retrieval corpus carries English, Hindi and Russian, so those three draw on translated source material rather than an English answer translated at the end.