AI TātparyaSign in

AI Usage & Beta Disclosure

Last updated: 29 September 2026

How this product was built

AI Tātparya's software — the website, the account system, the chat interface, and the underlying code — was built with substantial assistance from Claude (Anthropic's AI assistant), working under the direction and review of the founder. We're saying this plainly rather than leaving it undisclosed: every feature was specified, reviewed, and approved by a human before being shipped, but the code itself was frequently written by AI.

How the chat feature works

The chat assistant you talk to inside the app is also powered by Claude — a large language model made by Anthropic, not a human support agent. When you ask a question, we retrieve the actual text of relevant regulatory circulars from our archive (CERT-In, MeitY, RBI, IRDAI, SEBI) and give that text to Claude to base its answer on. This keeps answers grounded in real source documents rather than the model's general knowledge — but it does not make the model infallible.

This is a beta feature

The chat assistant is under active, ongoing development, not a finished product. We are continuously working on retrieval accuracy, answer quality, and reliability, and behavior you see today may change — hopefully for the better — as we improve it. Expect occasional rough edges: an answer that's less precise than it should be, a citation that's slightly off, or a response that takes longer than usual while we're still tuning performance.

Please don't share personal information in chat

Don't paste real people's personal information into the chat — names, PAN/Aadhaar numbers, account numbers, customer records, or anything else that identifies a specific individual. Ask about what a regulation requires, not about a specific person's or customer's situation. See our Privacy Notice for how long chat content is retained.

Text only — no file or image uploads

The chat accepts typed text only. You cannot upload or attach files, documents, spreadsheets or images, and the assistant does not read, open or analyse any. To ask about a passage, type or paste the text itself (without personal information). Each message is limited to 20,000 characters.

Tell us when it gets something wrong

If the assistant gives you an answer that's inaccurate, cites the wrong circular, or is otherwise unhelpful, we want to know — that's exactly the kind of feedback that improves it. Write to hello@aitatparya.com with what you asked and what went wrong.

Home