Legal
Sub-processors
Last updated: October 2, 2026
ChatbotIQ uses the following third-party service providers ("sub-processors") to deliver its services. All sub-processors are bound by data processing agreements ensuring appropriate safeguards for your data.
PostgreSQL, Redis and the vector database (Qdrant) are self-hosted on our own servers in the EU (Hetzner, Germany). Crawl4AI (web scraping engine) runs locally within our Docker containers. None of these constitutes a separate sub-processor.
| Provider | Purpose | Data Processed | Location | Training Opt‑out | Retention |
|---|---|---|---|---|---|
| Hetzner | Infrastructure hosting (servers, databases, vector storage) | All application data | EU (Germany) | N/A | Per our retention policy |
| Sentry | Error tracking & performance monitoring | Stack traces, browser metadata (no PII) | EU (Germany) | N/A | 90 days |
| Mistral | Chat LLM | Queries & document text | EU (France) | No training (per API terms). No opt-out needed. | Short-term (abuse monitoring) |
| Stripe | Payment processing | Name, email, payment info | US / EU | N/A | Per Stripe data retention |
| OpenAI | Embedding generation + Chat LLM | Queries & document text | US | No training (default since Mar 2023). No opt-out needed. | 30 days (abuse monitoring) |
| Anthropic | Chat LLM | Queries & document text | US | No training. No opt-out needed. | 30 days (safety monitoring) |
| Google Gemini | Chat LLM | Queries & document text | US | No training (paid API). No opt-out needed. | Per Google Cloud terms |
| Cohere | Search result reranking | Queries & document text excerpts | US | Training on our data disabled in account settings. | Per Cohere data policy |
| OAuth authentication | Email, name (via OAuth consent) | US | N/A | N/A | |
| Resend | Transactional email delivery | Email address, email content | US | N/A | Per Resend data policy |
Data Processing Notes
- LLM and reranking providers (OpenAI, Anthropic, Google, Mistral, Cohere): All are used via paid commercial APIs which are contractually excluded from model training. We do not use your personal data or user content to train general-purpose AI models.
- Sentry: Configured with
send_default_pii=False. No personally identifiable information is sent.
Changes to Sub-processors
We will update this page when sub-processors are added or removed. Material changes will be communicated via email to account holders.
Questions
If you have questions about our sub-processors or data processing practices, contact us at privacy@chatbotiq.eu.