Four capabilities. One governance runtime.
Most AI governance tools are dashboards. Constrix is infrastructure — inline, non-bypassable, and active before your AI executes. Here is what it does.
Every AI decision evaluated before it executes
Constrix sits inline in your AI pipeline. Before any agent acts, it passes through eight evaluation stages — schema check, risk classification, kill-switch, policy evaluation, conflict resolution, and cryptographic sealing. One of four decisions is returned in under 5ms: allow, restrict, deny, or kill.
Control what your AI agents are allowed to do
AI agents connect to tools and data sources via Model Context Protocol (MCP). Without governance, agents can call any tool with any parameters. Constrix evaluates every MCP tool call before it executes — approving, restricting, or blocking access based on your governance rules.
Know exactly which AI agent did what — and when
Every AI agent deployed through Constrix receives a cryptographic identity. Actions are attributed to a specific agent, session, and timestamp — not just "the AI system." Constrix also detects ghost agents: agents that are no longer active but can still act within an open session.
Turn your regulation into working governance rules — automatically
ComplyAI maps regulatory articles from EU AI Act, ISO 42001, NIST AI RMF, HIPAA, GDPR, and others to Rego governance policies that Constrix enforces inline. When an auditor asks for proof, ComplyAI generates a signed evidence package — not a report, not a log, but a cryptographically sealed record that the regulation was followed at execution time.