What data do AI features use?
AI features send the prompts and workspace content needed to answer — see Privacy for subprocessors.
Updated 2026-08-13
AI features send your prompt and the workspace content needed to answer — such as relevant messages, tasks, or notes — to the model that fulfills the request. FabricLoop does not use customer content to train models without explicit consent.
What is sent
- The text (or voice transcript) you submit in Ask, Voice AI, Loop Agent, or another agent.
- Workspace context the feature needs to respond — for example, a thread you asked to summarize, or tasks matching a query.
- Nothing extra from unrelated channels or files that the tool did not need for that request.
MCP connections are limited to the tools and workspace you approved on the consent screen. You can revoke a client at any time.
Where to read the full policy
- Open Privacy Policy — or Settings → Privacy in the app.
- See how we share data with AI model providers (Anthropic, OpenAI) when you use AI features.
- For the current sub-processor list, open the Security page. Enterprise customers can also email dpo@fabricloop.com.
Tips
- Owners and admins can limit agent spend and plan if they want tighter control over how often models are called.
- HIPAA is available on request through a Business Associate Agreement — contact us to discuss a BAA. It is not a self-serve checkbox.
Related
- Can I limit AI usage for my workspace?
- How do I connect MCP tools like Cursor, Claude, or ChatGPT?
- What is Ask?
- AI & Agents
Still stuck?
Contact us or browse all help topics.