Sub-processors
A short list, kept short deliberately. Every entry here is a party that can technically touch customer data, and each one exists because there is no responsible way to do that job ourselves today.
Last updated 1 August 2026
Current sub-processors
| Hetzner Online GmbH | Infrastructure hosting: compute, storage, network and backup targets for every customer Site. |
|---|---|
| Location | Germany and the Netherlands |
| Data accessible | All data held in a customer Site, at rest and in transit within the platform. |
| Model inference provider | Executing AI agent requests. Only the specific records needed to answer a request are transmitted, the provider is contractually prohibited from training on transmitted content, and nothing is retained after the response. |
| Location | European endpoints where the provider offers them |
| Data accessible | The narrow context of an individual agent request, for the duration of that request only. |
| Resend | Delivery of transactional email sent by the platform and by this website's contact form, where you have not configured your own mail server. |
| Location | European Union |
| Data accessible | Recipient address, subject and message content of the email being sent. |
What is deliberately not on this list
Most software companies have a long sub-processor page because they have assembled their product from other people's services. We build and operate the platform itself, so there is no analytics vendor holding your usage, no customer data platform, no third-party support desk with a view of your records, no session recorder and no advertising network.
- No product analytics provider receives customer data
- No CRM or marketing platform holds data from your Site
- No third-party support tool has access to your records
- No data warehouse outside our own infrastructure
- No backup provider outside Europe
How model inference works
When an agent answers a question, the platform assembles only the records needed for that specific request, within the permissions of the person who asked, and sends that narrow context for inference. The provider is bound by terms that prohibit training on transmitted content and prohibit retention beyond the request.
Nothing is indexed, no conversation containing your business data is stored, and there is no accumulating copy of your records anywhere in the inference path. Customers with stricter requirements can ask about restricting inference to specific endpoints, or about a deployment where inference runs entirely on infrastructure dedicated to them. Write to khadeem.smart@tynk-tech.nl and you will be told exactly which provider serves your Site.
Changes to this list
We give customers at least thirty days' written notice before adding or replacing a sub-processor. If you reasonably object on data protection grounds, you may terminate the affected part of the service without penalty, as set out in the data processing agreement.
Every sub-processor is bound by obligations no less protective than those we owe you, and we remain fully liable for what they do with your data.
Notification
To be added to the notification list for changes to this page, write to khadeem.smart@tynk-tech.nl. Site administrators are notified by default.
Questions about this document go to khadeem.smart@tynk-tech.nl. Trivena Technologies B.V. (i.o.) is established in the Netherlands, and these documents are governed by Dutch law.