Map

Overview

Once agents are discovered, SinzAI Map builds a complete topology of what data each agent can access. It classifies every data store by sensitivity, regulatory regime, and region — then cross-references agent access against these classifications to surface "toxic combinations" where unverified or low-trust agents can reach sensitive data.

Map operates continuously. As data stores are added, as agent scopes change, and as classifications are updated, the topology is recalculated in near real-time. A toxic combination alert fires the moment a dangerous access pattern emerges — not during the next audit cycle.

Data Store Classification

SinzAI classifies every connected data store across three dimensions:

Toxic Combination Alerts

A "toxic combination" occurs when an agent with insufficient trust or verification can reach data that its classification does not permit. SinzAI defines toxic combinations as rule triplets: (agent_trust_level, data_sensitivity, action_type).