ON2IT - Zero Trust Innovators

Select your region

Talk to us →
AI Endpoint Assessment

Your EDR is quiet. That's the problem.

Employees and their AI tools install agents, extensions and MCP servers your EDR doesn't see.

Any partner can install AES. We translate the findings to your Protect Surfaces, and our 24/7 GSOC can keep watching afterwards.

Request the free assessment →
67 of 100 slots left this quarter
  • Free
  • Alongside your current EDR
  • Prioritised report
Two weekson selected endpoints
Zero Trustper Protect Surface
The blind spot

What your EDR doesn't see

An EDR watches processes and trusts applications like Chrome, VS Code and Node.js. But whatever runs inside them, such as extensions, packages, MCP servers and agent skills, has no process of its own. To your EDR, it doesn't exist.

Inside trusted apps
Extensions and packages run inside software your EDR already trusts. A poisoned update doesn't stand out.
With real privileges
An AI agent works with the privileges of the employee who installed it. It reads mail, writes code and calls APIs. To your EDR, that is ordinary work.

The assessment maps this layer:

AI agents Browser and IDE extensions Packages MCP servers Agent skills AI models
Already happening

Attacks that stay under the radar

No exploit, no break-in, no alert. These attacks came in through the layer your EDR trusts.

npm + GitHub
Shai-Hulud
Infected npm packages harvested secrets from developer laptops and spread to thousands of GitHub repositories.
Browser extensions
Glassworm
Invisible code in developer tools spread itself into new software built with them.
IDE extension
TeamPCP
An update from a hijacked maintainer account stole Vault tokens, SSH keys and cloud metadata. Endpoint controls stayed quiet.
Agent skills
ClawHavoc
More than 800 malicious skills broke out of the agent and gave attackers the laptop's command line.

Source: Palo Alto Networks, Definitive Guide to Agentic Endpoint Security (2026); Unit 42.

What you get

Four questions. One report.

What is running?
A complete inventory
Every AI agent, extension, package, MCP server, skill and model on the endpoints in scope, per device and per user.
What is risky?
Risks, ranked
Assessed per item and per version: malicious code, vulnerable dependencies, a new or unknown publisher, excessive permissions.
What can an agent do?
Insight into privileges
Which tools touch your business data, and through which account? And which Protect Surfaces are within their reach?
What comes first?
A plan in three tracks
Fix now, set in policy, or later. Translated into your Zero Trust approach, so you know where to start.
How it works

Two weeks of looking

  1. 01
    Prepare

    You give access to your management tool (MDM), choose a selection of endpoints and name a contact person.

  2. 02
    Look

    AES runs with its own sensor alongside your existing EDR. Running Cortex XDR? Then Cortex XDR One puts XDR and AES in one sensor and one console.

  3. 03
    Discuss

    We go through the findings together and translate them into your Zero Trust approach: fix now, set in policy, or later.

Why ON2IT

Proven technology. A partner who makes it work.

The technology
Palo Alto Networks
  • IDC MarketScape: Leader in modern endpoint security (September 2026), pointing to its visibility and control over AI agents.
  • AV-Comparatives: Cortex XDR a Leader in the 2026 Endpoint Prevention and Response test, for the seventh year in a row.
  • SE Labs: AAA rating for Cortex XDR in the August 2026 ransomware test.
The partner
ON2IT
  • Since 2009 a Palo Alto Networks partner, Global MSS Partner of the Year in 2020.
  • Zero Trust: we translate the findings to your Protect Surfaces. What access is really needed?
  • 24/7 GSOC: for what comes next, with MDR Detect™ or MDR Prevent™.
FAQ

What you want to know

What does it cost?
Nothing. The assessment is free: two weeks of visibility into the AI layer of your endpoints and a prioritised report.
Do I have to replace my current EDR?
No. AES runs with its own sensor alongside your existing EDR. Running Cortex XDR? Then Cortex XDR One puts XDR and AES in one sensor and one console.
Why doesn't my EDR see this?
An EDR watches processes. Extensions, packages, MCP servers and skills run inside trusted applications without a process of their own. And an AI agent works with an employee's privileges, so to an EDR what it does looks like ordinary work.
What does the assessment look at?
AI agents, browser and IDE extensions, packages, MCP servers, agent skills and AI models on a selection of your endpoints, per device and per user.
What do I need?
Access to your management tool (MDM), a selection of endpoints and a contact person. Two weeks later we go through the findings together.
Start by looking

Know what runs on your endpoints

After that, you know what to fix first, and how it fits your Zero Trust approach.

Requesting takes a minute.We get back to you within two business days.

67 of 100 slots left this quarter
  • Free
  • Alongside your current EDR
  • Prioritised report
AI Endpoint Assessment
Request the free assessment
Takes a minute

By sending this form you agree to our privacy notice.