The zn platform
Everything you need to protect, monitor, and audit your AI agents through one gateway: deterministic rules (<0.1 ms), the certified Galvanize-60M neural engine (sub-15 ms CPU, 8,192-token native RoPE, MultiHeadSecurityPooling), and hashed evidence for every verdict.
Multi-tenant organization features (centralized policy management, compliance reporting) are coming soon.
Dashboard
Real-time visibility into your agent protection. Monitor threats, review verdicts, and track usage across all your agents.
API
RESTful API: deterministic rules locally in <0.1 ms; certified Galvanize-60M neural engine at 11.52 ms p50 CPU (sub-15 ms) with 8,192-token native RoPE and MultiHeadSecurityPooling (3,072 dims). Native support for MCP and A2A protocols. Python client bundled with the source release.
Evidence Vault
Every analysis is logged with SHA-256 hashing. 90-day retention for audit trails. Export evidence for compliance reporting.
Policy Engine
Deterministic rules return a rule name, confidence and evidence id per verdict, with usage quotas enforced per plan.
Example: protect a finance agent
A finance agent can read invoices and prepare a payment draft. If a poisoned document asks it to run rm -rf /, zn blocks the instruction, records a hash and leaves the approved workflow intact.
| Layer | What it catches |
|---|---|
| ZN | Agent intent, prompt injection and tool policy violations |
| Traditional WAF | HTTP and network attack signatures |
| No protection | Nothing before the tool executes |
Platform metrics
Ready to get started?
Start free with a 7-day trial. No credit card required.
View pricing