Electronic security for people who know what they are asking for.

A security system is not bought: it is maintained. And it is not only security — underneath there is a network that has to carry the video, drives that decide how many days of recording there really are, and someone who picks up the phone at three in the morning. We do all three here, because all three decide whether what was installed still works.

On installation day everything works. What decides whether it still works in three years is whether someone inspects it, whether the alarms are tested, whether the cameras that have stopped seeing get replaced and whether the plan is updated when the operation changes. No supplier can hold that up alone: it takes someone inside who wants it.

That is why the conversations that go well all start the same way. Someone who already knows what a false alarm at three in the morning costs, who has had a camera pointing at a useless spot for two years, who knows why their insurer asks about certification. With that person you can talk about the project in the first meeting.

And the ones that go badly also start the same way: with a list of cameras and the question of how much each one comes to. There the honest thing is to say that we are not the cheapest and we are not going to be, because half of what we charge is the project done beforehand and the maintenance done afterwards, and none of that shows up in a price per camera.

A badly designed installation does not throw an error. It switches on, it works, it passes commissioning, and nobody finds out there is a gap until the day something happens right in that gap. It is the most expensive failure there is in this trade, because it is discovered late and with the site already built.

The project is deciding what is protected and from what. Which zones, which entry points, which hours, what happens when someone forces a door at four in the morning and who receives that alert. And then what equipment is needed at each point: where each camera goes, at what height, at what angle, what can be seen from there at night and in the rain, which route the cable takes and what happens if the power goes.

It is also what makes the whole thing compliant. Local regulations, whatever applies outside Spain if the installation is abroad, and the documentation an inspection or an insurance company will ask for later. A system that works but cannot be documented is worth half of what it was bought for.

The field of view decides what can be done with that image for good. A camera ten degrees lower reads a number plate; the same camera ten degrees higher sees the roof of the car, and no software fixes that. That is why the angle, the height and the light are argued out in the project and not on the day of the install.

And here is what separates us from an ordinary installer: we know what you are going to ask of that image in two years' time. Because Infinity Neural is in the same house, and they are the ones who make cameras understand what they see. When we build an installation knowing that one day it will have to count people, detect PPE or read a number plate, that camera goes where it needs to go from day one.

It is not compulsory, and most of our installations do not have it. But it leaves the door open at no cost, which is the opposite of finding out in two years that analysing means going back up the poles.

Almost every installation we come across is run with cards that get lent out, fobs that get lost and a list in a spreadsheet that nobody has updated since the person who kept it left. It works until you have to answer the question of who went into that room on Thursday — and then it does not.

Serious access control is three things at once: the door opens for whoever should get in, it does not open for whoever should not, and both of those leave a record. The third is the one usually missing, and the only one that helps when you have to justify something to somebody outside.

It integrates with the rest, which is where you can tell whether the system was designed or bought in pieces: an access denied out of hours can bring up a camera, alert the alarm receiving centre and be recorded with its image, instead of being a line in a log nobody reads.

The problem with intruder detection is not detecting: it is the false alarm. A perimeter that alerts five times a night over a branch, a cat or the wind stops being a security system by the third week, because whoever is on duty learns to ignore it. And a system that gets ignored is worse than not having one, because it gives the feeling that something is covered.

So the work is in the sensitivity and in the zones: which detector goes at each point, which technology stands up to that environment, how it is split into sectors so that one zone under building work does not knock out the rest, and what gets verified before anything sounds.

And in video verification, which is what turns an alert into a decision: whoever is on duty looks at what set the alarm off and decides, instead of sending a patrol out blind or not sending one at all.

It is the only thing on this page where a failure does not cost money: it costs something else. A badly placed detector, an uncovered zone or an installation that has not been tested in two years is never noticed, until the one day it matters.

That is why it comes with its own regulations, its own documentation and its own inspection schedule, and why it is the part we negotiate on least. If a quotation comes in cutting corners here, better not to do it at all: it is the one line we would rather walk away from than sign half-done.

It is designed with the rest and not separately, because the day it goes off several things have to happen at once: alert people, open what has to open, close what has to close and know who is still inside. That is four systems talking to each other, and they only talk if somebody thought about them together.

Cameras do not generate office traffic. A person asks for something, receives it and sits still for ten minutes; a camera starts sending on the day it is switched on and never stops, and it is not one camera: it is all of them at once, towards the same place. A network designed for the first holds up the second on commissioning day and starts running tight the month cameras are added.

And when it runs tight it does not throw an error: it drops packets. What you see is jumpy images and gaps in the recording that nobody connects to the network, so the camera gets changed, the recorder gets changed and the problem stays exactly where it was.

We do the whole thing: cabling, fibre, switches, power over the network cable itself, separating the cameras from the network the company works on, and the link to the outside. With the calculation in front of us before the first cable goes in, which is while it is still a decision and not a building job.

A number of terabytes is not a number of days. The days of recording depend on how many cameras there are, how much detail each one has and how much moves in front of them, so the same drive that gave thirty days in winter with twelve cameras gives eighteen in summer with eighteen. Nobody changed it, and it is no longer true.

The rest goes the same way. A server sized for recording chokes the day six people open sixteen-camera mosaics. A backup unit that feeds the recorder and not the switch keeps a recorder running that is receiving nothing. And a closed cabinet in a room with no ventilation takes the drives with it in two summers.

All of that gets calculated: how much has to be kept and how much it really takes up, what machine is needed today and what headroom it leaves for tomorrow, how many minutes of power and in what order things shut down. And it is checked with the system running, not on the datasheet.

You call at three in the morning. The question is not whether there is a number: it is who picks up, whether they can get into the system at that moment and what happens when it cannot be fixed from a screen. Four very different services are written as the same line in a quotation, and the difference shows up once a year, all at once.

The other half of the job is that you should not have to call. A camera that has gone blind is still switched on and still sending an image: the recorder sees nothing odd, and someone finds out about the three missing weeks on the day they were needed. We check that every day and we open the alert ourselves.

And preventive maintenance comes with dates, a route and a report of what was looked at point by point. A one-line annual report saying «installation inspected, correct» costs the same as an inspection and is not an inspection.

A conference, a trade fair, a festival. It is not a small installation: it is another trade. You have to cover a site that did not exist a week ago, to deadlines that do not move, knowing that on opening day there is no second chance at anything. And then take it down without leaving a trace.

We do it because we are good at the part almost nobody wants: planning a system that has to be perfect on the first day and not on the third.

When one company does the installation and another does the analysis, the day something stops working the finger-pointing starts: the installer says the camera is fine and the software firm says the image is no good. Both can be right, and the problem is still there.

Here the person who chooses the angle knows what will have to be understood from there, and the person who understands it can ask for the camera to move ten degrees. There are not two suppliers: there is one party responsible, and they answer for both halves.