Data, privacy and retention
AI features may share relevant account, website, document, conversation and attachment data with third-party AI providers so they can complete the requested work.
What Silktide controls
- Which providers may process your data at all, chosen by your administrators on the AI screen.
- Conversation history lets the Agent continue an existing thread.
- Attachments are supplied to an AI provider when needed to complete your request.
- Access to account context and tools follows the signed-in user's permissions.
- Account administrators can opt out of optional Agent features, or withdraw a provider, at any time.
Which provider processes your data
Only the providers your account has enabled. Silktide currently offers OpenAI, Anthropic, SpaceXAI, Google, and Mistral. See What AI providers does Silktide support?. This is a consent, not a preference: Silktide cannot route your data to a provider you have not allowed, and neither can Silktide staff working in your account.
Where a provider processes is stated on the AI screen. Mistral processes in the EU. Google can process in the US or the EU. OpenAI, Anthropic, and SpaceXAI process in the US. That is the whole location question for these APIs: we send a request and get a response, and any short-lived logs live in the same place.
Withdrawing a provider stops future work reaching it immediately, including work already in progress.
How long are conversations stored?
Silktide stores Agent conversations so users can return to previous discussions. You can delete a conversation.
Each provider retains API request data for its own limited period, typically for abuse monitoring, under its own API terms. Ask your Silktide contact to confirm the retention terms for the providers you have enabled, or if you need to make a deletion request.
Is customer data used to train AI models?
No. The providers Silktide uses state that API inputs and outputs are not used to train or improve their models. This is a property of API access, and it is not affected by which provider you choose.
What data is sent to AI?
Only information needed to answer the question or complete the requested task is sent to the AI provider. Depending on the request, this may include relevant website results, issue details, page content, account configuration, conversation history, or an attachment you provide.
Minimise what you share
Only attach information needed for the task. Avoid placing secrets, credentials, or unrelated into prompts. The Agent can often retrieve permitted product context through tools without requiring you to paste it into a conversation.
Related controls
See the AI screen for your account's provider consents and the record of every change to them, AI models and providers for how a model is chosen and Ask and Edit mode for what the Agent may do.