Rudran

Privacy policy

Effective date: 2026-05-24

This Privacy Policy explains how Rudran Labs Inc., trading as Rudran ("we", "us", "our"), collects, uses, retains, and shares your personal data when you use our Service. Rudran Labs Inc. is the data controller for personal data processed about you. Its incorporation has been filed in Canada and the certificate is pending as of this version.

1. Data we collect

We collect only the data needed to run the Service:

  • Account data: email address, name, hashed password (Argon2id; we never store cleartext), 2FA configuration if enabled.
  • Sign-in metadata: IP address and device user-agent. Used for suspicious-sign-in detection and your sign-in history.
  • Operational data: the threads and documents you create, the run traces and chat transcripts an agent produces, the activity log of actions it takes, and any third-party accounts you explicitly connect for an agent to use.
  • Cookies: an essential session cookie for sign-in; an analytics cookie only if you grant consent via our banner.
  • Anonymized telemetry: error reports and performance metrics via Sentry (sampled, no PII), only if you consent. When an error occurs this can include a session replay — a reconstruction of what was on your screen in the moments before it — so we name it rather than let "error reports" quietly cover it.

2. Purpose of processing

We process your data to:

  • Provide the Service (authentication, lead handling, AI agent runs).
  • Protect your account (suspicious-sign-in alerts, abuse prevention).
  • Keep the Service working (anonymized error + performance telemetry). Never to train models.
  • Communicate with you (transactional emails; product updates only with consent).
  • Comply with legal obligations (audit trail, retention windows).

3. Retention

  • Account data: retained while your account is active. On self-initiated deletion, your account is removed from service immediately and a 30-day grace window applies. Being straight with you: the automated purge that would erase the underlying records is written but is not yet running on any schedule, so final erasure is a manual operation for us today. If you ask us to erase your data we will do it and confirm. Do not read this as an automatic timer — it is not one yet.
  • Sign-in metadata: kept for suspicious-sign-in detection and your sign-in history. Automatic expiry is not yet enabled, so treat this as retained until you ask us to remove it.
  • Audit log: retained (we keep this beyond your account deletion as required for security forensics + legal obligations).
  • AI worker activity: the per-action log expires after 90 days. Run traces and chat transcripts do not yet expire automatically.
  • Anonymized telemetry: 90 days, then auto-purged.

4. Third parties

We use the following sub-processors to run the Service. Each handles only the data needed for its function. Straight about the legal layer: where we hold an account with a provider, we are on that provider's standard published terms and whatever data-processing addendum it publishes. We have negotiated nothing bespoke — we are pre-launch, with no leverage or counsel to do so. And some entries below are not contractual relationships at all: they are public endpoints your browser or our server calls without an account. This line changes when that does.

  • Railway — application hosting.
  • MongoDB Atlas — database hosting.
  • Resend — transactional email delivery.
  • Stripe — payment processing for credit purchases and for monthly plans. Runs stop when your balance reaches zero, unless a plan renews it or you have switched on auto-reload.
  • Better Auth (self-hosted) — authentication. We run the Better Auth library against our own database; account and audit data stay within our infrastructure and are not sent to a third-party auth host.
  • Ollama Cloud— LLM inference (the AI worker's brain) for the pre-launch developer tier, currently running gpt-oss:120b. Rudran has no house model: which provider processes your prompts follows the model you pick per agent. We do not use your prompts or outputs to train models.
  • OpenRouter— model routing for the other models in the picker. It forwards your prompt to whichever upstream provider serves the model you chose, under that provider's own terms. Some free-tier upstreams may use prompt data for training, so we cannot extend a no-training guarantee to every model you might pick; if that matters for your data, choose a model whose provider you have checked.
  • Anthropic— LLM inference for Claude models. These reach Anthropic one of two ways: routed through OpenRouter above, or called directly by the agent runtime using an Anthropic key held in the agent's sandbox. Either way your prompt reaches Anthropic and is handled under their terms.
  • Composio— brokers an agent's authorized connections to your other tools. It sees only the accounts you explicitly connect.
  • Tavily— web search on an agent's behalf; receives the search query.
  • FAL — media generation. The customer-facing image abilities are not switched on yet, but a FAL credential is provisioned inside the agent sandbox, so we disclose it rather than assume nothing can reach it.
  • Modal — secure sandbox infrastructure for the AI worker runtime.
  • Google (favicon service) — our public showcase pages load site icons from Google, so Google sees the IP address of anyone viewing those pages. No account or identity of yours is attached.
  • Have I Been Pwned — checks a new password against known breach corpora when you sign up or change it. Your password never leaves our servers: only the first five characters of its hash are sent, so the service cannot learn which password was checked.
  • Sentry — error + performance monitoring.

We do not sell your personal data, and we run no advertising or analytics trackers. One caveat we will not hide behind that sentence: our public showcase pages load site icons from Google, so Google sees the IP address of anyone viewing them. Nothing identifying you is attached. Signed-in pages can also reach third parties two other ways we will not paper over: error monitoring, if you consent to it; and any image an agent or a document embeds from an outside host, which your browser fetches from that host like any web page.

5. Your GDPR rights

If you are in the EU, EEA, Switzerland, or the United Kingdom, you have the following rights under GDPR / UK-GDPR:

  • Right of access — request a copy of all personal data we hold about you. Use Settings → Your data → Download my data.
  • Right to rectification — correct inaccurate or incomplete data. Email + name from Settings → Email.
  • Right to erasure("right to be forgotten") — delete your account and all linked data via Settings → Danger Zone. 30-day grace window applies.
  • Right to data portability — receive your data in a machine-readable JSON format. Same path as access.
  • Right to restriction — request we pause processing your data. Email privacy@rudran.ai.
  • Right to object — object to processing based on our legitimate interests. Email privacy@rudran.ai.
  • Right to not be subject to automated decision-making — we use AI for lead-handling, but every action it takes is logged + can be reviewed; significant decisions are not made solely by algorithm.
  • Right to lodge a complaint with your local data protection authority.

6. International transfers

Our hosting, database and sandbox infrastructure runs in the United States, and several sub-processors (our LLM providers among them) are US-based — so if you are outside the US, your data is processed there. As section 4 says, we are on those providers' standard published terms, which is where any transfer mechanism, such as Standard Contractual Clauses, comes from. We have not negotiated our own, and we are not going to imply otherwise.

7. Security

We use industry-standard security practices: Argon2id password hashing, encryption in transit (TLS), encryption at rest (provider-managed), principle-of-least-privilege access controls, audit logging, and ongoing security reviews.

8. Changes to this policy

We may update this Policy. Material changes will be communicated via email at least 30 days before they take effect.

9. Contact

Privacy questions or to exercise any of the rights above, email privacy@rudran.ai.

This document is a pre-launch draft and has not been reviewed by counsel. It will be replaced with the final lawyer-reviewed version before paid customer onboarding.