AI Disclosure
How Pocket Pandit separates religious authority from generated conversation
1. The AI is not a religious authority
Pocket Pandit uses a large language model (LLM) for conversation. LLMs predict text; left alone, they can invent scripture citations, Sanskrit mantras, or planetary timings. We design Pocket Pandit so that the model explains and converses, while calculations and sacred texts come from governed sources.
Our principle
The conversation is flexible; the tradition is governed. The ritual is structured, and the model is not allowed to present invented material as authority.
2. Calculated vs. generated
- Panchang: Tithi, Vara, Nakshatra, Yoga, Karana.
- Sunrise, sunset, and moonrise timings.
- Chart calculations: Lagna, Bhavas, Vimshottari Dasha.
- Muhurtham scoring and dosha checks.
- Mantra texts and ritual step sequences, from reviewed sources.
- Conversational replies in supported languages.
- Explanations of symbolism and meaning.
- Answering your questions and interruptions during a ritual.
- Synthesised speech for spoken guidance.
3. Our design commitment against fabrication
We design the assistant to check requests and replies against a registry of attested sources. When it cannot find reliable textual support for a ritual, quotation, or rule, it is built to say so rather than invent one, for example:
“I could not establish this reliably from the sources available to me.”
AI systems can still make mistakes. Treat generated explanations as guidance, not scripture, and consult a qualified Pandit or scholar for important observances. Unresolved questions may be added to our research backlog for review.