QUESTIONS
Endpoint questions: answers on agentic endpoint security
SHORT ANSWER
Short answers to the questions buyers ask most about controlling AI agents on devices and in cloud workspaces, grouped by theme, with links to the page that goes deeper. Answers that mention scores are computed from the same data as the rankings.
By The Agent Desk · Agentic Endpoint Compare · Published 2026-10-11
Category basics
What is agentic endpoint security?
The control layer for AI agents that act with a person's access on laptops, desktops and cloud workspaces. It finds agents, sets rules on what they may do, and decides on single actions before they run. See lesson 1.
How is it different from EDR?
EDR watches for malware and intrusion. Agentic endpoint products decide what an AI agent may do, such as whether a tool call runs. They work alongside EDR, and several deploy through it.
What is a pre-tool call?
The moment after a model proposes a tool call and before the tool runs. It is the main point where an agent's action can be stopped. See lesson 2.
Why is blocking not enough?
A control that can only block gets switched off when it blocks real work. Asking a person, warning, masking or simulating first keep people working while still changing the outcome. See lesson 3.
Do agents in cloud workspaces need different controls?
Often. An agent on a laptop that calls cloud CLIs can be controlled on the device; an agent that runs in a cloud workspace needs coverage where it runs. See lesson 4.
What is an MCP server, and why does it matter here?
A program that gives an agent tools or data over the Model Context Protocol. Each MCP server an agent connects to widens what it can do, so approved-server lists are a common first rule.
Buying and pricing
Does any vendor publish prices?
None of the products we score publishes list prices on its own website. Noma Security and Lasso Security list annual contracts on AWS Marketplace ($200,000 and $20,000 per 12 months for the listings we read; marketplace list prices), and Harmonic Security publishes its tier names. Budget with vendor quotes.
Which products deploy without a new agent on the device?
On the pages we read, Bay deploys through existing EDR or MDM as an ephemeral binary, and Noma Security discovers endpoint agents through existing EDR or MDM with no new endpoint agent. Harmonic Security rolls out through Intune, JAMF, Kandji or Group Policy.
Should I buy from my EDR vendor?
If one platform covers your whole estate, its agent module is worth a serious look; Koi is part of Palo Alto Networks and Prompt Security part of SentinelOne. If your estate is mixed, weigh products that deploy through what you run. See lesson 6.
What should I ask a vendor with little public documentation?
Architecture, data handling, attestation reports, supported agents, failure behaviour, the method behind vendor-stated figures, pricing and references. See lesson 8.
Which products name the most coding agents?
Zenity and Ent each name 4 of the six agents we track. The full agent census lists every vendor.
Which products document the most verdict types?
Bay, Onyx Security and Neo each document 4 of six verdict types. The full verdict census lists every vendor.
The products
Which product ranks first?
On our published weights, Bay ranks first at 7.20 out of 10, followed by Neo at 6.85 and CrowdStrike Falcon Guardian at 6.25. See the rankings.
Where does Bay score lowest?
Bay is lowest or tied lowest on public proof, and trails at least one product on cloud and workspace reach, named agent coverage and public proof. See the Bay profile.
Does Bay replace EDR?
No. Bay states that it deploys through existing EDR or MDM. The EDR keeps handling malware and intrusion; Bay adds decisions about what AI agents may do.
Which product covers agents in cloud workspaces most fully?
Onyx Security, Zenity and Noma Security score highest on cloud and workspace reach, at 9 out of 10.
Who owns Koi and Prompt Security?
Palo Alto Networks completed its acquisition of Koi on 14 April 2026 and sells it as Cortex Agentic Endpoint Security. SentinelOne announced its agreement to acquire Prompt Security on 5 August 2025, and the deal was reported closed on 5 September 2025.
Which product fits a team that mostly uses Claude Code?
Several vendors name Claude Code: Bay, Neo, Zenity, Noma Security, Lasso Security, CrowdStrike Falcon Guardian, Harmonic Security, Ent, Koi (Palo Alto Networks), Prompt Security (SentinelOne) and Glow Security. Compare them in stack compare and ask each how deeply it controls Claude Code.
Which product is best for Microsoft, Salesforce or ServiceNow agents?
Zenity lists integrations with Microsoft 365 Copilot, Copilot Studio, Salesforce, Agentforce and ServiceNow among others. See the Zenity profile.
Is there an open-source option?
Lasso Security describes an open-source MCP gateway. The other products we score do not describe open-source components on the pages we read.
How this site scores
How are the scores calculated?
Each product is scored 0 to 10 on eight criteria; the total is the sum of score times weight divided by 100. Two criteria, verdict range and named agent coverage, are computed from our censuses. See Our scoring.
Did you test the products?
No. Scores come from desk research on public vendor pages read on 27 September and 11 October 2026. There was no hands-on testing and no vendor briefing.
How do I report an error?
Use the contact page with the topic Correction and a link to the public source. We check it and update the data.