It goes in in a day. And that has to be told with its three questions.

Ajax Systems is a European manufacturer that came in relatively recently and did it the other way round from everyone else: it started with how the system is lived with —the app, the deployment, the design of the kit— and worked up in capability from there. Today it has a wide range of radio detectors and sounders, panels in several sizes and also bus-wired devices for anyone who needs them.

You find it in retail, hospitality, offices, clinics, homes, residents' associations, branch offices and a great many installations that previously had nothing because cabling was unviable. And it is worth saying what it does exceptionally well: it deploys extremely fast, it is extended in five minutes and the app is the best in this group. Not «it has an app»: it is genuinely good, and a system the user understands is a system the user arms every night — half the installations that protect nothing are like that because nobody arms them.

At the centre there is a small panel acting as hub and as communicator. The detectors talk to it over a proprietary two-way radio protocol: the panel asks each device and knows whether it is alive, with how much battery and with what signal. It is encrypted, it hops frequency and it detects if somebody tries to knock out the radio with a jammer, which is the modern way of attacking a wireless system.

Two things that break the cliché. One: there is a line of bus-wired devices, so this system does not have to be all radio. Two: there are detectors that send a sequence of photographs of whatever set the alarm off, and that is verification by image inside the intruder detection system — it is not video surveillance, but for deciding whether to go out, it serves. Outwards, communication over the network and over the mobile network with more than one route, transmission to a receiving centre with the usual protocols, and an installer app for remote maintenance that is, bluntly, the most comfortable in this group.

Autonomy. The honest answer is not a number, it is a relationship: the devices last years on their batteries, but how long depends on how the supervision is configured. The more often the panel asks whether the detector is still alive, the sooner you know one has died and the less the batteries last; the less often, the longer they last and the later a dead point is detected. That interval is governed by the grade the installation calls for. And there is a logistical consequence quotations ignore: twenty devices are an annual inspection; two hundred are a calendar, a battery type in stock and somebody to look after it.

The real range. The published one is measured in open ground and is good for comparing, not for designing. The one that matters is the one in that building, with the racking full, the doors closed and the concrete slab between floors; an industrial unit with a steel frame and steel racking eats any manufacturer's radio. That is why we do not quote a wireless system without measuring the signal at the exact point of every detector, and we write down what margin each one has on handover day: that is what lets you know, a year later, whether the problem is the system or the pallet somebody has put in front of it.

And the grade, which is the one that decides projects. It is set by the risk, the activity and very often the insurance company, and at the upper end of that scale there are requirements for supervision, autonomy and resistance to sabotage that push you towards cable in particular zones or across the whole installation. Which grade each device is certified for and in what version is in its certificate with a date —that is where you look, not in a brochure and not on this page—, and which grade an installation meets depends on the whole build and is signed by whoever hands it over. If what the insurance asks for cannot be reached this way, that is said before signing, even if it means selling something else or selling nothing.

Better than many expect, and for a reason that is not the detection: it is the image. An alert that reaches your phone with a sequence of photographs of what set the detector off is dismissed or escalated in ten seconds, and that is what separates a system still in use the following year from one that gets ignored.

And what it does not fix, which is most of it: the fitting. A detector put where the signal reached instead of where the detection was needed. An infrared beam facing a west-facing metal shutter, which fires at dusk when the sheet cools in gusts. One in the blast from the air conditioning, which fires when the machine starts. An indoor one covering a terrace: cat, tarpaulin and wind. And the masking fault every morning at the same time, which is not a fault but something parked in front of it. The difference is that with photographs that is diagnosed in an afternoon, and it is fixed by changing the position, the lens or the angle, not the sensitivity. And the zone under building work is isolated on paper and with a date: in a system this easy to switch off from a phone, that discipline is more necessary, not less.

It is a separate decision, with its own contract and its own regulations, and it is worth settling before buying: which protocol is used, exactly what information reaches the receiving centre that is going to look after the installation, and whether it can work with the images, because that depends on their platform and not only on the panel. In Spain the receiving centre is obliged to verify the alert before passing it to the State security forces, so a system born with the image inside it fits that obligation well — provided the receiving centre can make use of it.

It fits very well what most buildings in this country actually are: the shop, the office, the clinic, the restaurant, the good house, the residents' association, the modest industrial unit, the branch. It fits where something has to be protected without building work and without closing the business for a week, and when the client is going to live with the system from their phone. And it fits especially well where things really are temporary: a building site, a unit rented for two years, an area that is going to be reconfigured.

It does not fit when the grade the insurance or the activity demands pushes you to cable across the whole installation, and that genuinely happens: there are sectors where the conversation ends there, and better before selling. Nor in the large installation with many areas, a supervised ring bus and a security department with an operator. And it does not fit if nobody is going to take charge of the batteries, which is the only maintenance this system really demands. Put the other way round, and it is fair: in the installations where it fits, it is hard to justify why you would fit anything else.

The client can run almost all of the day-to-day from the app, and does it without much teaching: named users with permissions, who arms which group, arming and disarming remotely, part-arming, schedules, alerts to whom and when, one-off isolation of a device, history with its photographs, and the battery and signal status of the whole system. It is more than the user of almost any classic panel can do, and it is a real advantage we are not going to play down.

What needs the installer is less than on the others, but it exists: adding devices in an installation where the grade matters, changing the supervision, the times or the confirmation logic, touching the transmission routes, and rechecking the signal after a refurbishment. And if it was handed over with a grade, any of those means the documentation has to be reviewed. The batteries are on the border: with twenty devices the client changes them if the type is written down for them and it is explained that opening the cover sets off the tamper; with two hundred it is a calendar in the contract.