Legal
Subprocessors
Last updated: 2026-10-14
The service providers that process personal data when Valnora & Elric runs Agent Serve for you, what each one does, and when it is involved.
The short version
- Agent Serve runs on Amazon Web Services in Frankfurt, Germany.
- AI model and service providers are involved only when you use the feature that needs them.
- Integrations and model accounts your organization connects itself are its own choice and are not listed here.
- Customers that host Agent Serve in their own cloud choose their own providers.
Who we are
Agent Serve is a product of Valnora & Elric Private Limited. In these documents, "Valnora & Elric", "we", "us" and "our" mean Valnora & Elric Private Limited.
| Entity | Registration | Registered office |
|---|---|---|
| Valnora & Elric Private Limited (India) | CIN U62099PN2025PTC245748 | A5, S. No. 16/2, 1st Floor, Swapnamandir Society, Deccan Gymkhana, Pune 411004, Maharashtra, India |
Which entity you deal with: if you are a customer, your contract is with the Valnora & Elric entity named in your order form or agreement. If no entity is named, it is Valnora & Elric Private Limited. For visitors to our website and people who contact us or request access, Valnora & Elric Private Limited is responsible for your personal data unless we tell you otherwise.
Our subprocessors
A subprocessor is a company we use to process personal data on behalf of our customers. Each one may use the data only to provide its service to us.
| Provider | What it does | Data involved | When |
|---|---|---|---|
| Amazon Web Services | Hosting in Frankfurt, Germany (eu-central-1): servers, database, cache, file storage, transactional email, secrets and logs. Amazon Bedrock for some AI models, which may run in other AWS regions. | All customer data and account data | Always |
| Anthropic | AI model provider (Claude models) | Prompts, files and context sent to the model, and its output | When a Claude model is used through Agent Serve’s own provider account |
| OpenAI | AI model provider | Prompts, files and context sent to the model, and its output | When an OpenAI model is used through Agent Serve’s own provider account |
| TypeSafe | Classifies each Chat message to choose a model and the tools it needs | Chat messages and attachment types | When automatic model routing in Chat is switched on |
| ElevenLabs | Speech-to-text and text-to-speech | Audio and text sent for conversion | When speech features are used |
| Daytona | Isolated sandboxes that run code for workflows and Chat | Code and the data passed to it | When code-execution sandboxes are enabled and used |
| AgentMail | Email inboxes for Chat and Relays | Emails sent to and from the inbox | When a workspace turns on an email inbox |
Apart from our hosting in Frankfurt, these providers may process data outside the European Union, including in the United States.
Providers your organization chooses
When your organization connects an integration, such as Slack, Google Workspace or HubSpot, or uses its own keys for a model provider, such as Google Gemini, data goes to that service under your organization’s own account and agreement. Those services are not our subprocessors.
Changes and questions
We update this page before a new subprocessor starts processing customer data. The date at the top shows the latest version. Questions: systems@valnoraelric.com. See also our Privacy Policy.