Managed Defender is where it starts. EDR, SIEM and an AI SOC come with it.
Defend is the detection and response module in The One Stack, the platform MSPs run their whole business on. It manages Microsoft Defender as the antivirus on Windows, and that is the starting point, not the whole product. Its own EDR sensor runs inside the one agent on Windows, macOS and Linux, a SIEM layer reads those detections alongside the logs you connect, and THEO works alerts as an AI SOC analyst while your technicians decide on containment.
Detection and response (Defend) is part of Security, one of the six systems of The One Stack.
What it does.
Managed Microsoft Defender, for antivirus
On Windows, Microsoft Defender is the antivirus and Defend manages it. You set one baseline for cloud protection, network protection, controlled folder access, signature updates and scans. In report-only mode, Defend records Defender’s health and its alerts and changes nothing. In managed mode, the agent applies the baseline and applies it again when a setting drifts. Defender’s own detections land in the same queue as everything else.
The EDR sensor is included
Defend’s own EDR sensor ships inside the one agent and reports process, file, network and sign-in activity. Detections map to known attack techniques. It takes the place of a separate EDR product, so there is no second security agent to buy, deploy or keep running next to it.
A SIEM layer on top
Sign-in and audit logs from Microsoft 365, Entra ID and the other identity, cloud and firewall sources you connect are collected and read with the endpoint detections. Network events from secure access (Edge) and tamper alerts from the monitoring agent join them. Alerts about the same device or the same person’s account are grouped into one incident by fixed rules, so the grouping does not depend on a model.
An AI SOC, with your technicians deciding
THEO does the first-line analyst work. Before a high or critical detection pages anyone, THEO checks what kind of device it is, what else fired on that device around the same time and how common the same activity is across your own devices. It says whether the alert looks like a real attack or needs a person, writes the summary and proposes a response. Containment waits for a technician.
Mac coverage on Apple’s Endpoint Security framework
On a Mac, the sensor is a system extension that uses Apple’s Endpoint Security framework. Apple granted the Endpoint Security entitlement the extension runs under, and the Mac agent is signed with a Developer ID and notarized by Apple. It reports process starts and exits, file changes, mounts and kernel extension loads. On macOS 13 and later, it raises a detection when the malware protection built into macOS finds or removes malware.
Linux coverage from the kernel
On Linux, the sensor reads process, file and network activity from the kernel, on x86-64 and ARM64 systems. On an older kernel it falls back to the audit subsystem, with less detail. Linux devices sit in the same endpoint list and the same incident queue as Windows and Mac.
Threat intelligence
Addresses, domains and file hashes seen on devices are matched against built-in threat feeds and the feeds you add. Domains are also checked against the reputation data that DNS filtering (Shield) keeps. A detection carries what is known about the indicator.
Response actions a technician approves
From the alert, a technician can ask the agent to isolate the device, stop a process or quarantine a file. The request travels as a signed command, and a device shows as isolated only after its agent confirms. Network isolation from the incident is available on Windows and Linux devices.
The queue starts with what needs a person
Incidents are ranked by priority, and the ones waiting on a decision sit at the top with the reason in plain words. Each response choice states what it does and whether it can be undone. One may be marked as recommended, and a technician approves it before anything runs.
Lookalike domains
On a recurring schedule, Defend looks for registered web addresses that imitate a client’s domain or a vendor the client relies on. Each match is listed with why it was flagged, and the client sees their own list in the portal. A message from a lookalike sender goes to a person as a ticket. The match alone does not quarantine it.
Inside detection and response.
More of the product, screen by screen.
Every open incident, ranked
Incidents from Windows, macOS and Linux devices and from Microsoft 365 sign-ins share one list, ordered by priority. Each row names the device or account and how many alerts were grouped into it.
Windows, Mac and Linux in one list
Every device running the sensor is in one endpoint list, whatever it runs. Each row shows whether the device is online, its agent version, its open threats and when its sensor last checked in.
Microsoft Defender, managed from one baseline
On Windows, Microsoft Defender is the antivirus and Defend manages its settings. In managed mode the agent applies the baseline and applies it again when a setting drifts. A device running another antivirus is only observed.
The logs behind the SIEM layer
Each log source you connect shows its health, when its last event arrived and what the last sweep stored. A source that has gone quiet is visible on this page.
How it connects.
Detection and response reads and writes the same record as the rest of the platform.
A detection opens a ticket
A critical or high detection opens a ticket with the details of the detection and the device on it.
A critical alert pages on-call
The page and the ticket go out together, and the result of the page is recorded on the action. The page title carries THEO’s read of the alert.
The sensor is part of the one agent
Defend loads as a module of the agent already on the device. Hostname, operating system and last-seen time come from the same device record as monitoring.
DNS filtering shares what it knows
A domain seen on a device is checked against the reputation data DNS filtering (Shield) keeps. A domain Shield rates as malware, phishing or a botnet raises a detection in Defend.
A ransomware alert leads to a restore
A technician picks a restore point taken before the encryption began and starts a scoped restore from the Defend console.
A serious incident gets a room
The incident’s timeline, playbook, client updates and evidence sit in one war room. Isolating a device from the room is a Defend request, and the device shows as isolated only after its agent confirms.
What THEO does here.
THEO investigates a high or critical detection before it pages anyone, writes a summary in plain language and proposes a response. Isolating a device and changing a user’s account always wait for a person.
Questions
Questions about detection and response
Is Defend just managed Microsoft Defender?
No. Managing Defender as the antivirus on Windows is where Defend starts. It also includes its own EDR sensor, a SIEM layer that reads the logs you connect alongside endpoint detections, THEO working alerts as an AI SOC analyst, and coverage for macOS and Linux, where the antivirus built into Windows does not exist.
Does Defend replace Microsoft Defender?
No. Microsoft Defender stays the antivirus on Windows, and Defend manages it. What Defend replaces is a separate EDR product, because its own EDR sensor is part of the agent.
Does it replace the EDR I have now?
Yes. Defend’s own sensor is the EDR. During the move the agent installs beside your current EDR or antivirus and leaves it alone, so you can go one client at a time. Running both is a step in the move, not how Defend is meant to run, and you remove the old product when you choose. While another antivirus is the active product, Defend does not turn Microsoft Defender on.
Is the AI SOC a staffed security operations center?
No. The AI SOC is THEO doing the first-line analyst work: investigating an alert, giving its read and proposing a response. There is no staffed security operations center and no managed response. Your technicians decide on containment, and a critical alert pages your own on-call rotation.
Who responds to an alert?
Your team does. An alert arrives as a ticket with a summary and a proposed response, and a technician approves any action on the device.
Is the SIEM layer a log search tool?
No. It collects the log sources you connect, raises detections from them and attaches the events to the device, the user and the incident they belong to. You read them there, in context, and not in a query console.
What does Defend do on a Mac?
The sensor is a system extension on Apple’s Endpoint Security framework, under the entitlement Apple granted for it, in a signed and notarized agent. Someone approves the extension and grants Full Disk Access on the Mac, and the device page says when a permission is still needed. Detections from the Mac land in the same queue as every other device.
Which operating systems does Defend cover?
Windows, macOS and Linux, through the one agent. Managed antivirus applies to Windows, where Microsoft Defender is built in. The EDR sensor, the SIEM layer and THEO cover all three.
Related
Keep going
Start
Run on One.
Start a trial and bring one client across, or book a walkthrough and follow a call down the line with us.