When IT leaders ask what AI-operated support looks like, the honest answer is: it looks like your Monday morning ticket queue having fewer tickets in it than it had on Friday. It looks like an infrastructure incident arriving with device context, recent change history, and affected users already attached. It looks like a new employee's laptop being ready on day one without anyone from IT manually coordinating the provisioning steps.
It does not look like a chatbot. It does not look like a feature toggle in your ITSM platform. And it does not maintain itself.
An AI-operated IT environment has three active layers.
Zendesk is where employees interact and agents work. It structures service requests, incidents, problems, and changes. The AI layer inside Zendesk handles triage, suggested responses, sentiment detection, and copilot assistance. Purpose-built for service delivery, handling high volumes of structured, repeatable requests. Exactly where AI agents perform best.
NinjaOne sees every device in the environment in real time: what it is, who has it, what is installed, whether it is healthy. When something changes or degrades, NinjaOne knows before the user does. That signal, without an operations layer connecting it to the service layer, goes nowhere useful.
L5 is the layer that makes the other two produce outcomes. L5 builds the integration logic that connects NinjaOne signals to Zendesk incidents, governs the AI agents that process those signals, maintains the knowledge base that deflection depends on, and runs a Drive every week to tune performance against the previous week's data.
"There is a caring and feeding to this AI world that people do not understand fully. You have to have a governance model, checks and balances, audits and controls."
Ron Mechling, CRO, L5[5]
AI-operated IT support means AI agents handle the repeatable work: routing, deflection, correlation, remediation. A human operator governs and tunes those agents on a weekly cadence. The agents do not run unsupervised. The operator does not wait for tickets to surface problems. Every week is a Drive with a defined output and a measured outcome.
Here is what a real AI-operated IT Monday looks like in an environment L5 operates.
6:00 AM: Overnight monitoring surfaces three alerts. NinjaOne has detected that fourteen laptops in the Chicago office have a failed Windows update from the weekend patch cycle. The L5-operated agent correlates the fourteen alerts into a single structured incident in Zendesk, pre-populates it with the affected device list, patch version, and failure reason, and assigns it to the endpoint management team. No one opened a ticket. No one manually correlated the alerts.
8:30 AM: A new employee starts. HR submitted her onboarding request in Zendesk two days ago. NinjaOne pulled a laptop from inventory, pushed software policy and access configuration overnight. Zendesk shows the onboarding ticket as complete. Her manager gets a notification. Her laptop is waiting at her desk. No IT coordinator was involved in the provisioning sequence after the initial workflow was triggered.[7]
10:15 AM: She cannot connect to the VPN. She types her issue into the Zendesk help center. The AI agent searches the knowledge base, finds the VPN client installation guide for her device type, presents it in the chat interface, and asks if it resolved the issue. It did. The ticket closes automatically. No human agent touched it.
"When we bring L5 into a deal, we win. They bring the ITSM best practices and pre-configured workflows that turn a platform sale into a production outcome. That is the difference between a customer who goes live and one who goes dark."
Spenser Mooney, Account Executive Employee Services, Zendesk[4]
2:00 PM: Three users in finance report they cannot access the ERP system. Each submits a separate ticket. The L5-operated agent detects the pattern: same application, same error code, same time window. The three tickets are promoted into a problem record, pre-populated with affected users, the application, the error, and a link to the last change that touched the ERP integration. The assigned engineer has full context before opening the record.
4:30 PM: A compliance check runs. NinjaOne scans every device for current antivirus signature versions. Eight devices are out of compliance. The L5-operated workflow generates a change request in Zendesk with affected devices, risk classification, and a proposed remediation window. The change is approved. NinjaOne executes the update overnight. By Tuesday morning, compliance shows 100%.
Monday's outcomes are not accidental. They are the result of the previous week's Drive. Every week, L5 reviews what the agents produced: deflection rates, incident correlation accuracy, false positives, escalation patterns. Then tunes the logic for the following week.
The knowledge base that deflected the VPN ticket was updated three weeks ago after the Drive showed a spike in VPN-related tickets the agent could not resolve. The agent identified the gap. L5 wrote the article. The deflection rate for VPN issues improved the following week. This is what an operated system does: it learns from its own performance data on a weekly cadence.
Most organizations that have purchased AI features in their ITSM platform have something different. Agents configured at go-live and not tuned since. A knowledge base populated during implementation and not updated as processes changed. Routing rules that worked when they were written and now produce a percentage of misrouted tickets that nobody tracks because nobody is watching.
"A lot of people have adopted AI and there is a lot of sprawl. You built an agent and walked away. What do you think is happening to the prompts that the agent is doing? Who is watching it?"
Pulkit Keshar, COO, L5[6]
The difference between AI features and AI operations: Features are capabilities that exist in the platform. Operations is the practice of governing those features: monitoring performance, tuning logic, updating knowledge sources, and measuring outcomes against defined targets. Features without operations produce agents that drift. L5 provides the operations layer that most ITSM deployments are missing.
AI-operated IT support is not a product you buy. It is a practice you maintain. The technology is a prerequisite, not the answer. What produces Monday morning outcomes is the operator running Tuesday's Drive.