Sub-processors
Every party with access to customer data is listed here. A sub-processor engaged without this list being updated would be a breach of our data processing agreement, so this is the authoritative set.
Infrastructure
| Sub-processor | Purpose | Location |
|---|---|---|
| Amazon Web Services | Platform hosting, compute, storage and managed databases | Customer selected region, including the United States and Singapore |
| OVHcloud | Platform hosting for customers requiring EU data residency | European Union, Singapore |
AI model providers
| Sub-processor | Purpose | Location |
|---|---|---|
| OpenAI | Language model used by the Kai AI agent to understand and draft repliesAPI data is not used to train OpenAI models. | United States |
| Anthropic | Language model used by the Kai AI agent to understand and draft repliesCommercial API data is not used to train Anthropic models. | United States |
| Gemini language model used by the Kai AI agentPaid API data is not used to train Google models. | United States |
AI and customer data
Kai, our AI agent, uses large language models from established providers to understand a message and draft a reply. Only the conversation content needed to answer, together with the knowledge base content the customer has published to Kai, is sent to the model provider. Model providers are sub-processors and are listed as such.
- Customer data sent through these providers' business APIs is not used to train their models, under the providers' own published terms, which are linked in the sub-processor list.
- AI features are optional. A tenant that does not enable Kai sends nothing to a model provider.
- Prompts and responses are processed to answer the request, not retained by us for model development.
- Customers requiring that no data reaches a third party model provider can run a private deployment.
Customers under contract are notified before a new sub-processor with access to customer data is engaged. To subscribe to those notices, email support@klink.cloud.