How Ishvaram writes, checks, and corrects what you read

Who writes what

Three different authors stand behind an Ishvaram page, and each is allowed to write only one kind of thing.

• The calculation engine writes every Vedic fact: sunrise and sunset, tithi, nakshatra, planetary longitudes, Lagna, Rahu Kaal and choghadiya windows, muhurat timings, Vimshottari dasha periods, Gun Milan scores, and dosha detection. It runs the Swiss Ephemeris with the exact parameters published on our methodology page, contains no language model, and makes no network calls. Generate your kundali.
• The festival calendar writes festival definitions and their published dates. A calibrated date carries two independent traditional authorities on record, and an automated check fails any release where the engine's own tithi resolution and a published festival date silently drift apart.
• The AI Jyotishi writes explanation only. It reads the chart the engine already computed and describes what those placements mean in the vocabulary of classical Jyotish. It is disclosed as an AI wherever it speaks, and it is never presented as a human astrologer.

What the AI is allowed to write

The AI may frame. It may not compute. That line is enforced in the software, not left to good intentions:

• A language model is never asked for a date, a time, a planetary position, a score, or a dasha period, and it is never allowed to emit one as a fact. Generated text keeps a placeholder where such a value belongs, and the value is filled in from the engine's own snapshot when the page is rendered.
• No AI checks another AI's facts. A generated body is scanned for stray facts before it is stored and again before it is published; a body that fails the scan is not published, and an input that cannot be read counts as a failure rather than a pass.
• Titles, descriptions, and social previews go through the same scan as body text, so a page cannot promise a date in its headline that its body never computed.
• Interpretation is anchored to the classical texts, chiefly Brihat Parashara Hora Shastra and Phaladeepika. Where lineages genuinely disagree, we publish the uncontested layer rather than one tradition's view dressed up as universal fact.

The same rule applies in the AI Jyotishi chat: the answers you read are framed around your computed kundali, and the AI is architecturally prevented from inventing a placement it was not given.

Who reviews

Ishvaram's Jyotish methodology is reviewed for traditional accuracy by Pandit Kuldeep Shastri, Vedic Astrologer & Content Reviewer. Practising Vedic astrologer (Jyotish), 26+ years, serving at Geeta Temple, Karnal. The public reviewer page carries a page-by-page ledger, and an entry appears there only after a completed human review is tied to the exact version of a page. Earlier review work is never backdated, and a page without a ledger entry makes no claim to one.

Editorial decisions about what Ishvaram publishes, what it declines to publish, and how it responds to a correction are made by the Ishvaram team described on the About page. The engine computes the facts, the shastras decide their meaning, the AI Jyotishi frames them in plain language, and an experienced human astrologer vouches for the approach.

Horoscope prose is guidance, not fact

On the rashifal, weekly and monthly horoscope, and transit pages, the planetary positions and dates are computed and reproducible. The reading written around them is a traditional interpretation offered for reflection and timing. It is not a prediction with a guarantee attached, and reasonable astrologers can read the same sky differently. Nothing on Ishvaram is medical, legal, financial, or psychological advice; our disclaimer says how to use the guidance responsibly, and any decision you take after reading it remains your own.

How corrections are handled

If a value on Ishvaram disagrees with a careful traditional Panchang, or a reading describes a placement your kundali does not contain, tell us through the contact page. A reported discrepancy is checked against the engine and the independent authorities named on the methodology page. A confirmed error is corrected at its single source, so every page that shows that value changes together, and an automated regression test pins the corrected value so the same mistake cannot quietly return. A wrong date under a sacred decision, whether a marriage muhurat, a naming ceremony, or a griha pravesh, is treated as an outage rather than a typo.

What we never do

• Present an AI as a human astrologer, or give it a fabricated name or lineage.
• Invent credentials, reviews, or review dates for anyone named on the site.
• Let a language model recall a planetary position or a festival date from memory.
• Publish a generated text that failed its fact scan, or backdate a correction.

See also the methodology page for the exact calculation parameters, the reviewer profile of Pandit Kuldeep Shastri, and our disclaimer. The privacy policy explains how we handle information you share.