Your CortexGraph workspace contains your company's most sensitive intellectual property — decisions, customer data, internal knowledge. It is treated as such.
No Training, No Telemetry
We do not train models on your workspace. We do not sell or share its contents. We do not log queries, primitives, or agent activity beyond what is required to operate the platform and provide auditability to the workspace owner.
Data Residency
Markdown sources, primitives, threads, and run records are stored in your workspace's isolated tenant. Access is scoped per actor, attributed to the human or agent that performed it, and gated by OAuth on every MCP connection.
LLM Provider Policies
CortexGraph itself is the context layer. The agents you connect — Cursor, ChatGPT, Claude, your own — speak to their underlying LLM providers directly under those providers' privacy and data policies. We do not proxy or store the model calls themselves.
For highly sensitive workspaces, we recommend agents wired to self-hosted or privacy-preserving model endpoints.
Website Analytics
This landing page (cortexgraph.ai) uses privacy-respecting analytics purely to monitor aggregate traffic. We do not collect personal identifiers, deploy tracking cookies, or sell visitor data.