Common questions
Does any data leave our machines?
No — Cortex runs entirely locally. Conversations, memory, file indexes, and RAG all live on the machine. The only outbound traffic is to the model providers you explicitly configure, and those can be fully local too.
What does “free core” actually include?
Everything Cortex does today: the WebUI, workspaces, memory, retrieval, controlled file access, and the expert panel. Enterprise adds services around the product — rollout, governance, support — not gated features.
How is this different from a cloud AI assistant?
Cortex is a composed runtime, not a hosted chatbot. Behavior is assembled from your workspace configuration and plugins, your data stays put, and compliance can audit each workspace independently.
What are the system requirements?
Windows, with PowerShell for the launch scripts. Model requirements depend on the providers you choose — hosted endpoints need almost nothing; local models need suitable hardware.
Can we add our own tools and experts?
Yes. Matbot remains the underlying runtime and plugin system — plugins can be added at runtime, and expert-specific files let you shape each expert’s knowledge per workspace.