Do you need 24/7 live monitoring, or are AI alerts enough?

What each one actually is

Every unit we deploy does the detection layer itself. On-board AI classifies what it sees, decides whether a person or a vehicle has entered a zone you defined, and sends an event with a clip to whoever is on the notification list. That happens with no human in the loop, it happens at the same speed at 3 a.m. as at noon, and it is included in the base rate on every unit.

Live monitoring adds people to the far end of that pipeline. A detection reaches a monitoring workflow where a human confirms what is actually happening, and then follows your escalation plan: talk-down over the speaker, a call to your contact list, a call to law enforcement, or a documented no-action for the raccoon. The hardware is identical. What you are buying is the response, not the seeing.

That distinction is the whole decision, and it is worth stating plainly because the industry blurs it deliberately. Some providers bundle monitoring into a single headline rate so the unit looks expensive and the response looks free. We price them separately — the unit with AI alerts is the base price, monitoring is an add-on from +$499 per month — because a site that already has someone on call should not be paying for a service it will never use.

The 2 a.m. question decides it

Detection is only worth what the response is worth. A perfect alert delivered to a phone on silent, in a house where nobody has agreed to be responsible for it, has the same practical value as no alert at all — with the added cost of feeling protected. Before comparing packages, find out honestly who is going to answer, whether they have agreed to it, and what they are authorised to do next.

The honest test is not whether someone could answer. It is what happened the last three times something went off in the middle of the night. Teams consistently overestimate their own after-hours responsiveness, and the pattern is predictable: the first week everyone checks every alert, and by week six the notifications are muted. If that is the realistic trajectory for your team, buy the monitoring and stop pretending.

When AI alerts alone are the right answer

A large share of sites genuinely do not need monitoring, and we would rather say so than sell it. Alerts alone are the right call when someone is already awake and accountable at night, when the site has a superintendent or owner who lives nearby and will drive out, or when the practical purpose of the system is deterrence and evidence rather than live intervention.

It is also the right call where the realistic response is not urgent. Plenty of properties are protected adequately by a recorded, classified, time-stamped account of what happened, reviewed the next morning. If nobody would drive out at 2 a.m. regardless of who called them, paying for a human to make that call adds cost without changing an outcome.

• Someone on the team is genuinely on call at night and has agreed to it
• The site is low-consequence enough that morning review is an acceptable response
• The unit is there mainly to deter and to document, not to interrupt
• You are early in a deployment and do not yet know your real alert volume

When monitoring earns the add-on

Monitoring pays for itself where an interruption in the moment changes the outcome, and where nobody on your side will reliably provide it. The clearest cases are sites with an active, ongoing theft problem rather than a theoretical one; properties where the loss from a single successful visit is large relative to the monthly cost; and unstaffed sites with no local person who can plausibly be woken up.

The second class is about accountability rather than economics. Where a lender, an insurer, a client or a board wants documented verified response — not just recorded footage — a monitoring workflow produces exactly that: a verified event, a documented action, and a timestamped record of both. That paperwork is sometimes the actual deliverable, and no amount of phone alerting substitutes for it.

It also changes the conversation with law enforcement. A verified report describing a person cutting a fence, observed live, is a materially different call from an unverified alarm activation. Where response priority matters to you, human verification is the mechanism that produces it.

• An active theft pattern, not a hypothetical risk
• High single-incident loss relative to the monthly add-on
• No credible person on call after hours
• A lender, insurer or client requiring documented verified response

What live monitoring cannot do

Monitoring is a response service, not a physical presence, and the limits should be stated before you buy rather than discovered during an incident. It cannot physically stop anyone. It cannot make police arrive faster than their own dispatch priorities allow. It cannot see what the cameras cannot see, which means a bad placement plan cannot be fixed by adding humans at the other end of it.

It also does not substitute for the coverage decisions that come first. A unit sited where it misses the actual approach produces verified reports about the wrong part of the property. Get the placement and the detection zones right, then decide about monitoring — in that order, because the second cannot repair the first.

And it is a recurring cost that scales with nothing. Unlike the unit itself, the monitoring add-on does not become more efficient over a long deployment; it costs the same in month twelve as in month one. That is an argument for reviewing it periodically rather than treating it as permanent once bought.

The sequencing that saves money

For most sites the cheapest correct path is to start without monitoring and let real data decide. Deploy the unit, tune the detection zones and schedules for the first two weeks — that tuning period is where most nuisance alerts get eliminated — and then look at what the site actually produced: how many genuine after-hours events, at what hours, and how many of them anybody responded to.

That gives you an evidence-based answer instead of a guess. A site producing two real overnight events a month that someone handled fine does not need a monitoring package. A site producing repeated events that nobody acted on has just proved the case for one, and the add-on can be turned on without changing hardware, moving the unit, or renegotiating anything.

The reverse also holds and is worth saying: monitoring can be dropped. If an active problem resolves, or the site phase changes, that is a monthly cost you can stop carrying. Treat it as a dial rather than a decision made once at signing.

How to compare monitoring offers

Monitoring is where security pricing gets least transparent, so compare on mechanics rather than headline rate. Ask what a human actually does when an event fires, and in what order. Ask whether talk-down is included or extra. Ask what the escalation list looks like and who maintains it. Ask what happens on a false alarm and whether you are billed for it. Ask what documentation you receive afterward, and how long it is retained.

Then ask the awkward one: is monitoring optional? A provider that will not sell you the unit without it is not offering a security decision, it is offering a bundle. Our published add-on rate exists so that the comparison is possible at all — the unit is $995 per month, monitoring is from +$499 per month, and you can run those two numbers against your own site before anyone talks to you.

Frequently asked questions