Privacy

Mitram is a new service. These notes describe what we collect, why, where it lives, and how to delete it. They are plain-English; if you want a formal policy, ask and we'll write one.

What we store

• Sign-in identity: email + Supabase user UUID. Supabase handles authentication on our behalf — your password (if you use one) is theirs to store, not ours.
• Profiles you create: name, kind (self / family / friend / partner / other), and optional notes. Each profile may have one chart attached.
• Chart inputs: birth date, time, timezone, latitude, longitude, and place name. This is required for the ephemeris compute.
• Computed chart data: lagna, planet positions, dasha periods, yogas, and so on — all derived from the inputs above.
• Decisions you make: the question, the agent's answer, the per-system trace, your reported outcome (if any).
• Agent conversation memory: durable user-memory entries (preferences, recent decisions) the agent uses to keep context across sessions.
• API usage: per-call token counts and cost so we can enforce daily caps and bill ourselves correctly.
• Payments: if you buy credits or subscribe, our payment processor handles the card details — we only store a payment reference (a Dodo customer / subscription id) and your credit ledger and purchase history. We never see or store your card number.

Where it lives

All transactional data sits in a single Supabase Postgres database. The knowledge graph (classical-source citations, planetary tables) is a static SQLite file bundled with the backend container — it has no per-user data.

How your data improves Mitram

Mitram improves by measuring itself against real outcomes. When you report how a decision turned out, that outcome — together with your chart data — may be used internally to calibrate and improve the system's accuracy. You consent to this when you onboard. Two hard rules apply:

• Your name is never sent to the AI. The chart context we pass to Anthropic is astrological only (lagna, nakshatras, yogas, dashas) — names and other identifiers are stripped before the call.
• This is for our own measurement and improvement — we don't sell it, share it with third parties, or use it to train shared external models.

What we don't do

• We don't sell or share your data with third parties.
• We don't share your profiles or charts with other users — every read scopes by your user ID at the database layer.
• We don't use your conversations to train shared models. The Anthropic API call sends your message + chart context to their inference servers; per their terms it's not used for training.
• We don't track you across the web. We use PostHog for product analytics (which features get used — signups, decisions run, outcomes reported) and Sentry for crash reporting. Both receive only your interactions and an opaque account id — never your name, birth data, or chart contents (autocapture and session recording are switched off precisely so birth details can't leak).

Keeping signups clean

To stop bots and abuse at sign-in, we use Cloudflare Turnstile (a privacy-friendly captcha — Cloudflare processes the challenge to confirm you're human; it doesn't track you across the web) and we screen out disposable / throwaway email domains at signup. These checks run on the sign-in step only and don't touch your chart or reading data.

Deleting your data

Email email.piyush@gmail.com with the subject “mitram: delete my account” and the email you signed in with. We'll wipe your row and all attached profiles, charts, decisions, predictions, outcomes, memory, and usage records. Self-serve deletion is on the roadmap (docs/12-backlog.md) but not shipped yet.

Service caveats

Mitram is a new service. Things may break, change, or be deleted without notice. Don't rely on Mitram for time-sensitive irreversible decisions while the service is new — see Terms.

Governing law

Mitram is operated by an individual based in India; this policy is governed by the laws of India.

← Back home · Terms · Refunds