The vetted third-party providers we use to run TextSight, what each does, and the data involved. We keep this list current and update it when it changes.
To provide TextSight, we rely on a small set of trusted third-party providers ("subprocessors") for hosting, payments, AI processing, email, analytics, and support. Each is used for a specific purpose and only receives the data needed for that purpose. We choose established providers with their own security and privacy programs.
| Provider | Purpose | Data involved | Region |
|---|---|---|---|
| DigitalOcean | Application hosting & APIs | Account & application data | EU / US |
| Vercel | Marketing & tools site hosting | Usage / request data | Global edge |
| Hetzner | Self-hosted AI-detection model servers | Text submitted for detection (transient) | EU (Germany) |
| Cloudflare | DNS, TLS, CDN & edge security | Request metadata | Global edge |
| Anthropic (Claude) | Generative / rewriting features | Text submitted to those tools | US |
| Sightengine | Image & audio AI detection | Uploaded image / audio files | EU |
| Stripe | Payments & subscription billing | Billing & payment data | US / EU |
| Managed PostgreSQL | Primary database | Account & application data | EU / US |
| Postmark | Transactional email | Email address, message content | US |
| Google Analytics (GA4) | Website analytics | Usage data, cookies | Global |
| Meta (Pixel) | Advertising measurement (marketing site) | Usage data, cookies | Global |
| PostHog | Product analytics | Usage / event data | EU / US |
| Sentry | Error monitoring | Diagnostic / error data | US |
| Crisp | Live chat & support | Messages you send us | EU |
Regions reflect where each provider typically processes data; some operate global infrastructure. Where data leaves your region, transfers rely on appropriate safeguards (see our GDPR notice).
The text you submit for AI detection is processed by our own self-hosted model (Hetzner, EU). Text you send to generative or rewriting tools is processed by Anthropic; images and audio you upload for detection are processed by Sightengine. None of these are used to train TextSight's models, and free scans are not stored. See Content Handling.
We may add or replace subprocessors as the service evolves. When we make a material change, we will update this page and the "Last updated" date above. Business and Enterprise customers who have a data processing agreement with us can request advance notice of new subprocessors — see our DPA.