Z
ztna-success
POC progress dashboard
← POCs

test Cato Customer

Cato Networks · owner dror@globadots.com · status scoping

Customer share link
https://ztna-success.globaldots.net/s/sh_9imvcizuh257jnzyqgki
Overall
0 / 1
criteria passed
POC phase
0 / 1
what the customer validates
Project kickoff
0 / 0
runs post-signing
POC view 0/1 Project kickoff 0/0 All 0/1

LLM API-based observability and guardrails

security Not started POC nice to have

For LLM calls the business itself makes (application backends, internal agents, AI copilots), a vendor-provided layer sits between the app and the model provider — observing prompts/responses, enforcing guardrails, and surfacing usage and cost in a dashboard.

Test procedure — Cato Networks · Cato AIM Security — AI observability and guardrails for internal LLM-using apps (acquired 2026)
Vendor pack default (read-only):
1. In CMA, provision AIM Security and create an AI observability endpoint for the target LLM provider.
2. Point the test app's LLM base URL at the AIM endpoint (or integrate the AIM SDK).
3. Send a handful of prompts from the app; confirm requests appear in AIM's observability view with per-request metadata.
4. Configure a guardrail or rate-limit policy; confirm enforcement on a policy-violating prompt.
Pass threshold: LLM calls are visible per request in AIM's dashboard; a guardrail or rate-limit fires on policy violation.
+ New custom criterion + Add criteria from library