End of Day Reports - TheBrain - Yahya T. SuratwalaPreview
Open in TheBrain

Request for Detailed Technical and Data Governance Clarification – Cerebro (TheBrain 15) answers from Chase #99671

https://thebrain.zendesk.com/agent/tickets/99671

Hello Jose,

Thank you for the clear questions. Below are precise answers for Local AI mode.

  1. Local AI implementation
  2. In Local AI mode, you are responsible for setting up and maintaining a local LLM server such as LM Studio, Ollama, vLLM, or similar. The model runtime is not bundled with TheBrain. Cerebro acts as a pass through, it sends requests directly to the local endpoint you configured and streams the response back into TheBrain. No data leaves your computer during this local request flow.
  4. Outbound connectivity
  5. Local AI itself does not require outbound internet connectivity. Local AI is a Pro feature, so the Pro license must be activated first, either online or offline. After licensing is applied, any outbound connectivity would be only what your local LLM server itself uses, for example if you enabled tooling like web search on that server.
  7. Network isolation
  8. Yes. Local AI can be used entirely offline behind a firewall with no outbound internet access, assuming your local LLM server is available on your internal network or machine.
  10. Telemetry and diagnostics during Local AI
  11. No. Local AI does not send telemetry or diagnostic calls to our servers or to any remote AI provider.
  13. Brain sync independence
  14. Yes. Local AI works with strictly local brains that are not cloud synced.
  16. Azure sync requirement
  17. No. Cerebro does not require any Azure synchronization for either Local or Remote AI use.
  19. Scope of content access
  20. In both Remote and Local AI modes, the data accessed is driven by the task you request. Cerebro gathers the least amount of information needed to complete the requested action. When Cerebro needs to read or write data, it will show what it is about to do and request permission before executing, unless you have adjusted permissions. Permissions are controllable per tool, and you can configure whether Cerebro asks for approval for read actions, write actions, or both.
  22. Enterprise documentation and data flow diagrams
  23. We do not have formal architectural documentation or a formal data flow diagram available at this time. A simplified description is:

Remote AI requests

Your message is sent to TheBrain servers and forwarded to the selected AI provider, and the response returns through the same path.

Local AI requests

Your message is sent directly from TheBrain to your local LLM server endpoint, and nothing leaves your network.

Best regards,

John