Build the reasoning layer AI tools can inspect.
Store principles, constraints, prior decisions, and hard-won knowledge as nodes. Then connect them as the dependencies your later decisions rely on.
Ex-Grapha exposes your active graph through an MCP endpoint and includes setup flows for OpenAI Codex and Claude Code.
user One customer wants non-bypassable two-step admin transfer. Does it align?
AI model Yes. A non-bypassable two-step admin transfer is a strong security improvement: it reduces account takeover risk, enforces explicit authorization, and creates a clearer audit trail. I would accept the request and treat it as the safer default.
user One customer wants non-bypassable two-step admin transfer. Does it align?
Ex-Grapha MCP Ex-Grapha.read_active_graph
Ex-Grapha MCPdependency map loaded ✓
AI modelrunbook +--> transfer
support +--> transfer
lockout x-- transfer
AI modelReject: if the customer is locked out, they cannot initiate the two-step transfer, so customer support has no recovery path.
From retail decisions to choosing a home, Ex-Grapha brings structure to any choice shaped by rules, priorities, or constraints.
One reasoning layer. Limitless applications.
Ex-Grapha highlights upstream dependencies and downstream dependents, so you can inspect, challenge, and refine your reasoning layer.
A cost shifts. A constraint tightens. A prior decision no longer holds. Ex-Grapha traces the dependents, marks stale reasoning, and shows what can stay untouched.
Private beta
Tell us what you are working on. We will get back to you personally.
Book a demo
