Privacy Policy

Who this covers

Two groups of people appear in our systems. Founders, who sign up and run validations. And authors of public Reddit and YouTube posts that our pipeline reads as evidence. The protections below are written for both.

Founder data

We process your account, run, and wallet data to deliver the service you signed up for. You can export your data, correct it, pause processing, opt out of marketing, and request deletion.

• Access & portability: export runs, reports, and wallet history in a machine-readable format.
• Deletion: an account-delete request purges your personal data across our stores within 30 days; only anonymized, non-identifying statistics may be retained.
• Rectification: edit your profile at any time from account settings.
• Objection: pause runs and opt out of marketing communications.

People in the evidence

We analyze public posts under a legitimate-interest basis, with a documented assessment that weighs necessity against impact. Personal identifiers are removed before anything is stored: handles are one-way hashed with a per-account salt, names and places are stripped by a named-entity pass, and a final cleanup pass catches the rest. Raw scraped text is never persisted — only the anonymized form reaches our database.

Anyone whose post was analyzed can request erasure, even if they are not a customer. We propagate that deletion across our database, object storage, search indices, and any cited quotes in reports within 30 days. We also poll the source platforms for upstream deletions and propagate those automatically.

Retention

Reports are kept as your deliverable until you ask us to remove them. Raw scraped content is purged on a schedule and never sold or licensed. Operational logs are scrubbed of personal data and kept short-term. Payment records are kept as long as tax law requires.

Where data lives & who processes it

V1 hosts in a single US region. EU residents can use the service; the cross-border transfer terms are disclosed here and Standard Contractual Clauses are in place with vendors that process EU data.

LLM vendors (Anthropic, OpenAI) operate under signed data-processing agreements with Zero Data Retention enabled and an explicit no-training-on-customer-data clause. Only already-anonymized payloads are sent to them.

Sensitive categories

Threads that look like they contain special-category information are dropped by default before any analysis. A thread is only ever retained if an operator explicitly approves it with a logged reason.