Astrix and Fortinet: Combining Network Visibility With Identity Context for AI Agent Security

Astrix and Fortinet are announcing a joint integration that combines FortiGate network telemetry with Astrix’s identity context layer to give enterprise security teams continuous visibility into AI agent activity. This post covers what each product contributes, how the integration works, and what security teams can actually do with it.

Astrix and Fortinet are announcing a joint integration for AI agent discovery and security across the enterprise.FortiGateNGFW sees what agents are connecting to. Astrix identifies who those agents are, what credentials they’re operating under, what they can access, and whether any of that activity represents risk. Together, they close a visibility gap that neither product covers alone.

The Challenge

AI agents leave traces in two places: the network traffic they generate and the credentials they use to authenticate against enterprise systems.

Most security tools see one or the other. Network telemetry without identity context tells you something connected. It doesn’t tell you it’s an agent a developer ran locally in Cursor last Tuesday, operating under a service account with admin access to production, with no named owner anywhere on record. Identity monitoring without network telemetry misses shadow agents entirely, because those agents were never registered on any platform.

That gap, between what’s visible on the network and what’s traceable through identity, is where most AI agent risk accumulates.

What Each Side Brings

FortiGate NGFW is Fortinet’s enterprise firewall and network security platform, deployed across perimeters in some of the largest organizations in the world. It delivers application-aware network traffic inspection across enterprise perimeters. Deep packet analysis, SSL inspection, application control, policy enforcement at the network layer. It captures traffic that no platform integration or endpoint agent ever sees, because it sits in the path of everything.

Astrix secures the identity layer that AI agents operate through. Every agent that reaches an enterprise system authenticates using a non-human identity: an OAuth app, service account, API key, or PAT. Astrix monitors those NHIs continuously across cloud, identity providers, SaaS, and DevOps tools, mapping each one to the agent behind it, the permissions it holds, the resources it can reach, and the human accountable for it.

How the Integration Works

FortiGate NGFW is configured to forward permitted HTTPS traffic logs to the Astrix platform via syslog. No new sensors. No endpoint changes. The telemetry is already being collected.

Astrix ingests those logs and runs them through its identity layer. Network traffic becomes identity context: which agent, which credentials, what it can reach, when it was first seen, and whether any of it violates policy. Discoveries are continuously updated as agent activity changes.

What Security Teams Get

The integration delivers five specific capabilities:

• Continuous shadow AI discovery. Agents deployed without security approval don’t appear on any platform dashboard. When an employee runs Claude, Cursor, or Windsurf on a corporate device, FortiGate logs the outbound traffic and Astrix surfaces the agent with full context, before ungoverned tools become incidents.
• MCP server inventory. Most organizations have no visibility into which MCP servers are active in their environment. FortiGate captures the connection. Astrix identifies the server, the device, the timeline, and whether the server is official, unofficial, or deprecated by its maintainer.
• Network-to-identity correlation. Traffic that previously had no identity context now does. Each discovery is mapped to the NHI behind it, the permissions it carries, and the resources it can reach.
• Risk assessment, not just detection. Astrix enriches each discovery with risk findings based on organizational policies, access scope, and blast radius. An agent in active use with write access to production gets flagged differently than a read-only integration.
• Auditable inventory. Every discovered agent and MCP server is logged with activity history, ownership status, and policy findings, so compliance and audit reviews don’t require manual reconstruction.

Why This Matters Now

Agents don’t just read data. They write, delete, and execute across systems using credentials that carry real enterprise permissions. The window between an agent getting deployed and security knowing it exists is where most of the risk builds up. Closing that window requires seeing the network and the identity layer at the same time. That’s what the Astrix and Fortinet integration is built to ensure.