A finance manager receives a convincing request to change a supplier’s bank details. Minutes earlier, the same employee’s cloud account logged in from an unfamiliar device. The email filter, identity service, and endpoint tool each noticed something, yet the small IT team saw three separate alerts rather than one developing incident.
AI Managed security closes that gap. Basically, it does the following:
- Sort telemetry
- Connect suspicious events
- Guide or automate response.
To do this, it uses machine learning and behavioral analysis. It also increasingly uses generative or agentic AI.
The managed part matters just as much. Someone must review what the models produce. They must also tune detection logic and decide whether disabling an account is justified.
1. Fortinet
Fortinet combines AI-assisted security operations with network, endpoint, cloud, and identity controls. It does so through its Security Fabric.
FortiAI capabilities cover the following:
- Alert triage
- Investigation
- Threat hunting
- Automation
- AI application visibility
- Protection against data leakage into large language models.
Its managed portfolio also includes SOC-as-a-Service and Managed FortiGate Service options.
So, Fortinet’s AI Managed security platform is worth examining when a company already runs FortiGate appliances or wants network telemetry to feed directly into security operations.
But don’t treat ecosystem alignment as proof of fit. Ask the provider to demonstrate detection across your –
- Microsoft 365 tenant
- Third-party endpoints
- Custom applications
- Non-Fortinet cloud services.
Also define which automated actions need approval. Fast isolation is helpful. But an incorrectly disabled production account isn’t.
2. Sophos
Sophos MDR provides 24/7 detection and response across –
- Endpoint
- Identity
- Cloud
- Network
- Business application data.
Its service combines autonomous investigation with human oversight. Sophos supports more than 500 security and IT integrations. So, customers can choose response arrangements that give the provider more or less authority.
This model suits businesses that want managed operations. Sophos also supplies –
- AI-generated case summaries
- Natural-language searches
- Command-line analysis.
Some AI functions assist investigations. They don’t directly perform remediation. Buyers should verify this distinction during evaluation.
3. Barracuda Networks
Barracuda Managed XDR gathers signals from –
- Endpoints
- Servers
- Networks
- Cloud services
- SaaS platforms
- Identity systems.
Automated detection works alongside a 24/7 security operations center. Meanwhile, its detection rules map activity to the MITRE ATT@CK framework.
In some cases, email and account compromise sit high on the risk register. For instance, a midsize professional-services firm may care more about suspicious mailbox rules and stolen Microsoft 365 sessions. They might not require highly specialized malware research.
Broad monitoring also comes with a cost. More data sources can create onboarding work. Response quality also depends on clean integrations. During a trial, deliberately break one connector. See whether anyone notices!
4. Zscaler
Zscaler takes an inline, zero-trust approach. This approach is toward users, devices, applications, data, and AI traffic. Its AI security capabilities cover –
- Application discovery
- Sensitive-data controls
- AI asset management
- Governance of model
- Agent interactions.
The managed-services question needs careful handling. This is because Zscaler is operated with a service partner. It does not have a conventional MDR package. For instance, Infosys and HCLTech pair Zscaler technology with managed zero-trust and AI-assisted security services.
5. Microsoft
Microsoft Security Copilot supports the following:
- Incident investigation
- Threat hunting
- Posture management
- Intelligence gathering
- Natural-language analysis.
It works with Microsoft Defender XDR and Sentinel. It also works with Entra, Intune, and third-party services including ServiceNow and Jamf.
Some businesses are already deep in Microsoft 365 and Azure. That proximity can reduce context switching. Still, Security Copilot alone isn’t a replacement for a managed SOC.
For instance, a company might need 24/7 human ownership. They must evaluate it alongside Microsoft security products and a managed service provider. Then, they must specify which incidents the provider may contain without waiting for internal approval.
Also scrutinize data access. Review the following:
- Permissions
- Plugin scope
- Audit logs
- Retention
- The material used to generate AI summaries.
6. IBM
IBM QRadar MDR uses AI-powered automation and machine learning with human analysis. This way, it provides the following:
- Continuous endpoint detection
- Alert investigation
- Threat hunting
IBM Managed Security Services can –
- Terminate malicious processes
- Remove files
- Apply blocking policies
- Isolate endpoints.
The broader QRadar suite spans SIEM and endpoint detection. It also includes behavior analytics, orchestration, and network activity. That depth may suit a larger organization.
It can also be heavier than a smaller firm needs. More control brings tuning, policy, and integration work. There’s a fair case for choosing a narrower MDR service if your environment is mostly cloud applications and standard employee laptops.
7. Arctic Wolf
Arctic Wolf’s MDR service collects telemetry from –
- Networks
- Endpoints
- Cloud environments.
Then, it blends automated analysis with 24/7 monitoring and a named security team. Its Aurora Agentic SOC keeps humans involved in response while the Concierge Security model adds remediation guidance and ongoing risk reviews.
This is a plausible fit for a business that already owns several security tools but lacks the people to operate them consistently. Still, log ingestion doesn’t guarantee detection coverage. Arctic Wolf’s documentation notes that some sources may lack complete parsing or rules, so buyers should obtain a source-by-source coverage map before signing.
Compare Operating Models, Not AI Labels
| Brand | Practical fit | Issue to test |
| Fortinet | Network-heavy or existing Security Fabric environments | Coverage beyond Fortinet products |
| Sophos | Lean teams wanting provider-led MDR | Response authority and escalation |
| Barracuda Networks | Email, identity, and cloud-focused businesses | Connector health and tool coverage |
| Zscaler | Zero-trust access and AI usage governance | Division of duties with service partners |
| Microsoft | Microsoft 365 and Azure-centered operations | Managed ownership and AI permissions |
| IBM | Hybrid enterprises with mature procedures | Complexity and administration workload |
| Arctic Wolf | Mixed-tool environments needing operational help | Parsing and detection coverage by source |
The Best Fit Is the Service That Can Act Responsibly
AI can shrink an alert queue and draft an investigation summary. It can also spot patterns that tired analysts miss. Moreover, it also produces confident mistakes. What still remains part of the product includes –
- Human accountability
- Transparent evidence
- Rehearsed escalation paths.
However, they don’t appear in the model description.
The right AI Managed security service should match your current tools and operating hours. They must also consider your regulatory duties and appetite for automated containment.
More features does not compensate for vague ownership during an attack.

Jeremy H. has opinions about tech news and innovations. Informed ones, backed by real experience — but opinions nonetheless, and they doesn't try to disguise them as neutral observation. They thinks a lot of what gets written about Tech News and Innovations, Tech Product Reviews, Practical Software Tips is either too cautious to be useful or too confident to be credible, and they's work tends to sit deliberately in the space between those two failure modes.
Reading Roys's pieces, you get the sense of someone who has thought about this stuff seriously and arrived at actual conclusions — not just collected a range of perspectives and declined to pick one. That can be uncomfortable when they lands on something you disagree with. It's also why the writing is worth engaging with. Roys isn't interested in telling people what they want to hear. They is interested in telling them what they actually thinks, with enough reasoning behind it that you can push back if you want to. That kind of intellectual honesty is rarer than it should be.
What Roys is best at is the moment when a familiar topic reveals something unexpected — when the conventional wisdom turns out to be slightly off, or when a small shift in framing changes everything. They finds those moments consistently, which is why they's work tends to generate real discussion rather than just passive agreement.

