Current sub-processors

Nine vendors. No surprises.

Most of these are foundational infrastructure. A handful are the AI models we measure on your behalf — we use their public APIs and never embed customer data in training calls.

Vendor Purpose Region What data is shared
Cloudflare cloudflare.com Edge compute, CDN, WAF, DDoS protection, R2 storage, D1 database, Workers KV. Global edge All customer requests transit the edge. Storage is U.S.-region by default.
OpenAI openai.com ChatGPT API queries for consensus scoring (GPT-4, GPT-4o). US Generated scan queries only. No customer PII. Zero-retention header set.
Google AI (Gemini) ai.google.dev Gemini API queries for consensus scoring. US Generated scan queries only. No customer PII. Data-not-used-for-training flag set.
Anthropic anthropic.com Claude API queries for consensus scoring. US Generated scan queries only. No customer PII. Workspace-level retention disabled.
Perplexity perplexity.ai Perplexity API queries for consensus scoring with web search. US Generated scan queries only. No customer PII.
PlanetScale planetscale.com Primary relational database for accounts, scans, reports, billing references. US Account data, scan metadata, report indexes. Encrypted at rest with AES-256.
Stripe stripe.com Payment processing, subscription management, invoicing. Global Billing email, card token, subscription state. We never see card numbers.
Resend resend.com Transactional and notification email delivery. US Recipient email, subject, body of system-generated emails.
AWS S3 aws.amazon.com Long-term cold storage for archived reports and audit logs. US-East Archived scan artifacts after 90 days. Encrypted with per-customer KMS keys.
i

How we add or change vendors.

Before adding a new sub-processor, we publish a notice on this page and email customers on monitoring plans at least 30 days in advance. Enterprise customers with signed DPAs receive direct notice. If you have objections, write to privacy@misquoted.ai — we'll work with you on alternatives where feasible.