How RankReply AI Protects Agency & Client Data
Last updated: October 25, 2026. Architectural transparency around data minimization, paid-tier AI processing, multi-tenant database isolation, and healthcare/legal writing guardrails.
1. Paid-Tier Gemini API (Zero Model Training)
RankReply exclusively calls the commercial paid tier of the Google Gemini API on our Edge servers. Customer inputs, Brand Vault configurations, and generated replies are never used to train foundation models.
2. Deterministic HIPAA & ABA Guardrail Engine
Every generated or manually edited reply runs through deterministic hard & soft checks across 1–5★ ratings — blocking patient/visit confirmation, echoed clinical terms, case/settlement references, and banned words before Copy unlocks.
3. Supabase RLS Multi-Tenant Isolation & Minimization
Workspace records are isolated via PostgreSQL Row-Level Security (RLS). Public review text is processed transiently in memory; at most a 140-character excerpt (review_excerpt) is saved in reply history.
4. Creem Merchant-of-Record PCI-DSS Compliance
Subscription payments, tax calculation, and invoicing are processed by Creem (Merchant of Record). RankReply AI never collects, transmits, or stores payment card numbers on our servers.
Zero Client OAuth & HIPAA Scope
Unlike legacy profile automation tools, the RankReply Chrome Extension operates directly inside your agency's authenticated browser session on business.google.com and Google Maps — eliminating the need to store client Google OAuth refresh tokens on third-party servers.
Important HIPAA Scope Notice: RankReply helps you avoid disclosing protected health information in public replies. We do not store full review text, we are not a HIPAA Business Associate, and you should never paste confidential patient medical charts into the app. Agencies and practice managers remain responsible for reviewing every reply prior to publication.
Infrastructure & Subprocessors
| Subprocessor | Purpose | Region / Safeguards |
|---|---|---|
| Supabase | Magic Link Auth & PostgreSQL Database (RLS) | US East (AWS) · Encrypted at rest & in transit |
| Cloudflare | Edge Hosting, Workers, KV Rate Limiting & Turnstile | Global Edge Network · TLS 1.3 + HSTS |
| Google Gemini API | Commercial Paid-Tier Reply Generation | Zero model training on API inputs |
| Google Maps Platform | Places API (New) for Competitor Audits | United States / Global |
| Creem | Merchant of Record (Billing, Taxes, Invoicing) | PCI-DSS Compliant Payment Processing |
| PostHog | Product Analytics | US Cloud · Zero review text or reviewer PII captured |
Data Retention & Deletion
- Reply History: Retained until you delete locations or request workspace erasure. Only a 140-character excerpt is stored.
- Extension Review Harvests: Temporary audit harvests expire automatically 24 hours after upload.
- 48-Hour Erasure SLA: Email [email protected] to permanently erase your workspace within 48 hours.
5. Responsible Disclosure & Security Contact
Have a security questionnaire for your agency or discovered a potential vulnerability? Contact our engineering security desk directly — we acknowledge all reports within 24 hours.