Keep your data on your own machines without turning your team into an IT department. We specify the hardware, build and install it, secure and back it up, and keep it running, with one named person who knows your setup. Getting private AI working for a week is the easy part. We take on the years after it, and the managed IT around it.
Nobody can tell you what hardware you actually need, so you either overspend or outgrow it in a year.
It works on somebody’s laptop. Getting it running properly on a machine in your building is a different job.
It gets set up once, by one person, and nothing about how it works is written down anywhere.
Models, drivers and the OS underneath all move. Deferred until something breaks is not a maintenance plan.
Nobody can say when they were last tested, as opposed to last run. Those are different questions.
When it does go wrong, it is not clear whose problem it is, so it becomes nobody’s.
We specify, build and look after the machines that on-premise AI runs on, and the ordinary IT around them: monitoring, maintenance, updates, backups and business server setup. If you are keeping your data in your own building, this is what actually makes that true rather than aspirational.
Perpetual Technologies is run by Samuel Edwards, from Southend-on-Sea. No account manager, no handover to somebody reading your setup for the first time while you wait. Pricing is fixed and transparent, and we explain what we are doing in plain language rather than in acronyms.
Data on your own hardware is an engineering property, not a policy. Somebody has to own the hardware for it to be true.
This service exists because of the product, and it is honest to say so. NeuralVault runs on hardware you own, and "hardware you own" is a promise somebody has to actually keep: the machine has to be specified, built, secured and maintained by someone, or the privacy argument is just words on a website.
That work is this page. The product is software you own and run. Everything around it is a job, and this is the job.
NeuralVault Free is a public download today: Windows only, and a beta under active development. During the Beta everyone has Personal, and it becomes a monthly subscription in the next few weeks, when the Stable version is released. Business is on request. The infrastructure work does not depend on any of it: on-premise AI runs on hardware whatever software you put on top.
Our managed IT work sits around on-premise AI deployments rather than being sold as a standalone contract. If you want infrastructure looked after and AI is somewhere in the picture, that is exactly what this is. If you are purely after a general IT helpdesk with no AI involved, we are probably not the closest fit, and we will tell you that on the call rather than after the invoice.
It depends entirely on the models you intend to run and how many people run them at once, which is why we will not quote a spec on a web page. A single developer running local models has very different requirements from a company of fifty sharing one machine. Working that out is the first thing we do, before anything is bought.
Usually, yes. Plenty of businesses already have a server doing very little. Whether it is suitable depends on what you want to run on it, and finding that out costs you a conversation rather than a purchase order.
Southend-on-Sea, Essex. We work with businesses across the UK and internationally, and hardware work in particular often wants somebody physically present. In person when it matters, fully remote when that is faster.
There is no public price list, because a deployment is scoped rather than picked off a shelf. Pricing is fixed and transparent once we know what you actually need, and the consultation that gets us there is free.
Book a free 30-minute call. We will look at what you are running now, what you want to run, and whether the hardware you have can take it. Straight answers, no hard sell, no obligation.