1. Script automation
Scheduled scripts and scripted monitors run on a fixed condition. No judgment, no context, no decision-making — just "if X, run Y." This is most RMM "automation" today.
AI RMM · What to demand from one
AI RMM: what it is, and what it should do for you
An AI RMM is an RMM whose operations are performed by governed AI — not an RMM with a chatbot bolted on top of a dashboard. This page defines the category, lays out what to demand from any vendor claiming it, and shows exactly how Breeze implements each requirement.
Definition
What an AI RMM actually is
Most tools calling themselves "AI RMM" have added a chat window on top of an existing dashboard. You still read the alert, still decide what to do, still click the button. That is an assistant, not an AI RMM.
A true AI RMM performs the operation itself: it investigates the alert, correlates the signals, and takes or proposes the fix — governed by a risk tier and logged for audit. The test is simple: when something goes wrong on the fleet, does the AI do the work, or does it just describe the work for a human to do? If it's the latter, you're buying a chatbot with an RMM attached, not an AI RMM.
The spectrum
From scripts to a managed AI team
"AI RMM" gets used for very different levels of capability. These are categories, not specific products — know which one you're actually evaluating.
2. AI assistant
A chat layer bolted onto the RMM that can answer questions or summarize data. Useful for lookups, but it does not act on the fleet — a person still reads the answer and does the work.
3. AI operator
AI with real tool access to the fleet: it can investigate an alert, correlate signals, and take action, classified by a risk tier and logged. This is the built-in Breeze operator, free in every deployment.
4. Managed AI team
The operator, supervised and tuned by the people who built it: someone reviews its ticket conversations, verifies resolutions held, and continuously improves it as your fleet changes. This is Breeze Managed AI Ops.
The buyer's checklist
What to demand from any AI RMM
Whatever vendor you're evaluating, these five things are non-negotiable. If a vendor can't show you all five, the "AI" is unsupervised — regardless of what the pitch deck says.
- 1
A risk-tiered governance model
Every action the AI can take should be classified into tiers, not left to the model's judgment alone. Reads should always be free. Low-risk, reversible actions should auto-execute with a log. Impactful actions should wait for a human. Destructive actions should be blocked outright.
- 2
A complete audit trail
Every tool call the AI makes on your fleet should be logged: what it looked at, what it proposed, what it executed, and who approved it. If you cannot reconstruct what the AI did after the fact, it is not governed — it is just running.
- 3
Human approval on impactful actions
Reboots outside a maintenance window, domain-controller changes, agent uninstalls, and anything else with real blast radius should require a person to say yes before it happens. An AI RMM without an approval gate is a liability, not a feature.
- 4
Supervision of the AI itself
Someone should be watching what the AI does on real tickets and real fleets, checking that its resolutions actually held, and tuning it when they do not. Governance controls what the AI is allowed to do; supervision checks whether it is doing that well.
- 5
The RMM works without the AI
If the AI layer goes down or you turn it off, the underlying RMM (monitoring, patching, remote access) should keep functioning. AI should multiply what your team can do, not be a single point of failure for core operations.
How Breeze implements it
The checklist, built in
Breeze is our answer to the checklist above. Full disclosure: we built it, so take the specifics and hold every AI RMM vendor to the same standard.
Risk engine, four tiers
Tier 1 reads run free — the operator can query device state, alerts, and inventory with no gate. Tier 2 low-risk actions (clearing a cache, restarting a stalled service) auto-execute and are logged. Tier 3 impactful actions (patching a domain controller, rebooting outside a maintenance window) wait for your approval. Tier 4 is blocked outright, full stop. You set the tier boundaries; the operator follows them.
How the risk engine works →Approvals, not assumptions
When an action lands in Tier 3, the operator drafts the request and stops. Nothing executes until a human approves it. The approval flow shows what will happen, on which device, and why the operator classified it that way — not a black box, a specific proposal you accept or reject.
See the AI assistant →Supervised chats, on the managed tier
On Breeze Managed AI Ops, we read the AI's ticket conversations, verify the resolutions actually held, and tune the agents with you as your fleet and clients change. You never have to become an AI expert to trust what it is doing — that supervision is the product, not an add-on.
Explore Managed AI Ops →Full audit trail, always on
Every tool call the operator makes — read, action, or blocked attempt — is logged with the device, the classification, and the outcome. The trail exists whether you self-host for free or run the managed team; it is enforced at the RMM level, not the AI level.
View all 64 modules →The RMM stands on its own
Breeze is fully functional with the AI layer disabled: monitoring, patching, remote access, scripting, and tickets all run without it. The operator is a layer on top of the platform, not a dependency — self-hosters can run the entire RMM free with no AI key configured at all, and switch the operator on later without redeploying.
Run the open source platform →Ticketing, quotes, and invoicing are built in too. Breeze is RMM and PSA in one platform.
Frequently asked questions
Answers to the questions people ask most often when evaluating an AI RMM.
Is an AI RMM safe on client fleets?
It is safe when the governance sits at the RMM level, not the AI level: reads free, low-risk actions auto-logged, impactful actions gated on human approval, destructive actions blocked. Ask any AI RMM vendor to show you the approval flow, not a highlight reel. If they cannot show a real approval screen with a real risk classification, the "AI" is unsupervised, whatever the marketing says.
What's the difference between an AI RMM and RPA or scripting?
Scripts and RPA execute a fixed sequence of steps on a fixed trigger — reliable, but with no judgment. An AI RMM investigates first: it correlates signals across the fleet, figures out what is actually wrong, and proposes or takes an action based on that reasoning, governed by a risk tier. Scripting answers "run this when X happens." An AI operator answers "what is actually wrong, and what should we do about it."
Do I need to be an AI expert to run an AI RMM?
No, and any AI RMM that requires that is asking you to take on a job you didn't sign up for. The built-in Breeze operator is governed by the risk engine, not by how well you can prompt it. On Managed AI Ops, we read the AI's conversations and tune it with you, so the expertise lives with the vendor, not with your team.
What does an AI RMM cost?
The built-in AI operator is free in every Breeze deployment (self-hosted or cloud), governed by the same risk engine either way. Breeze Managed AI Ops — the supervised team that also works your queue — is billed to your fleet and workload; it starts with a call, not a checkout, and the honest anchor is that it costs a fraction of the technician hire it replaces.
Can an AI RMM take destructive actions on my fleet?
Only within limits you define, if it is built correctly. In Breeze, Tier 4 actions are blocked outright regardless of what the AI recommends, and Tier 3 impactful actions cannot execute without approval. If a vendor's AI RMM can take irreversible action with no tier system and no approval gate, that is a design flaw, not a feature.
Is an AI RMM the same as an RMM with a chatbot?
No. A chatbot answers questions; an AI RMM acts on the fleet with real tool access, governed by risk tiers and logged for audit. A chat window that can only summarize a dashboard is an assistant bolted onto an RMM, not an RMM whose operations are actually performed by AI.
See the governed AI operator work a real fleet
Breeze's AI operator is built into every deployment, free, governed by the four-tier risk engine described above. Book a call to see the approval flow live, or deploy it yourself today.
Related reading