You bought the AI.
Nobody taught anyone to use it.
Most AI programmes do not fail on the model. They fail because the tool arrived, a link went round, and everyone was left to work it out between other jobs. This is the training half: a system your people work through, with videos, tasks and real problems they have to solve using an AI, built per department and ending in a qualification they earned rather than attended. It is in development and not yet on sale.
Six things that turn up in almost every business that has bought AI and not taught anybody to use it. If more than two of these are familiar, the licence is not what is wrong.
The rollout was a link in an email and a fifteen-minute demo. That is distribution, not training.
Asked badly, an agent does not stop and check. It guesses, then corrects its own guess, on your money.
It produces a clean paragraph with the wrong number in it, and nobody on the team knows what that looks like.
Some people stopped using it and did not say so, because admitting it is admitting they could not make it work.
Usage went up and nobody can point at what it bought. Spend is measurable, value is not being measured.
You cannot say who is genuinely competent with it and who is bluffing, so you cannot delegate anything real.
The skill nobody has is not clicking the right button. It is being able to say what you want clearly enough that a machine can do it, and then being able to tell whether what came back is true. Both are teachable, and almost nobody has been taught either.
So the course is built around the work rather than the interface: how to brief a job, how to check the answer before it reaches a customer, how to get your own methods and material into the tool so it stops giving generic replies, and where the money actually goes when you use it badly.
And it is a system rather than a session. People log in and work through it: a video and a written explanation of the concept, then a set of tasks where they have to solve a real problem using an AI and show what they got. Nobody passes by watching. The qualification at the end is a record of work somebody actually did, which is the only version of a certificate worth anything to the person signing it off.
Plenty of people will sell your team an AI course. Almost nobody will install the thing, teach your staff to use it on their own work, and then tell you which of them can.
The point of training is not that people have heard of the tool. It is that a specific person can be handed a specific job and trusted to do it well. These are the four things the course is built to leave behind.
They are deliberately capabilities rather than topics. "Covered prompt engineering" is not an outcome. "Can brief a job and catch a wrong answer before a client sees it" is.
It will be priced per seat, or per department where a whole team goes through it together, and it splits into a cheap general course everybody sits and a dearer specialist course built on your own systems. It will be available either alongside a NeuralVault deployment or entirely on its own, because most of what it teaches applies to whichever AI tools you already run. How that is put together is set out further down; there is no figure to quote yet and we would rather say so than invent one.
The service itself is under development and is not on sale. There is no date and no syllabus to send you, and nothing in an automation, integration or deployment engagement today depends on it. It is listed here because it is being built and because people ask for it, not because it can be bought.
Nobody learns one of these tools by being shown it. They learn by being given a real problem, trying it, getting it wrong somewhere it does not matter, and going round again next week. So it is a system your people log into rather than a session somebody delivers, and it runs alongside the job instead of stopping it.
Every round is the same four steps. Learn one concept from a short video and a written explanation, use it that week on work they were going to do anyway, hand in what they got, and go again with the next concept. The loop is the point: this is not a course somebody finishes on a Friday and forgets, it is a habit they build while the work carries on.
The qualification falls out of that record rather than out of a test at the end. By the time somebody has been round it enough times, you are not taking anybody’s word for what they can do, because the work is sitting there.
One concept at a time: a short video showing it done, a written explanation to go back to, then the tasks. Nothing is an hour long, because the design assumes people are fitting this between other work.
A real problem of the kind their department actually gets, solved using an AI and handed in. Not a quiz about what a token is. They do the thing once, badly, somewhere it does not matter.
Nobody comes off the job for this. What somebody picks up on the Monday is in use on the Tuesday, on their own work, which means the training starts paying back before it is finished rather than after.
A person qualifies on the work they actually submitted, so what you get is a record of what somebody did rather than confirmation that they were in the room. That is the entire reason it is worth certifying.
Fully current on how these tools work and what they cannot do, holding a qualification, and able to be handed real work with an AI and trusted to get it right. Not aware of AI. Able to use it.
Training a workforce once and then hiring three people is how a rollout quietly goes backwards. New starters come to us and go through the same course, so the standard holds instead of decaying with every intake.
Putting an entire company through the deep material is a waste of money, because most people do not need it. So it splits in two: a general course everybody sits, and a specialist course built on your own systems for the handful of people who go further with it, which in practice means your developers and whoever else is going to carry this.
The two are sequential rather than alternatives. Everybody starts on the general course, and the people going further go on to the specialist one afterwards, on top of what they already have rather than instead of it.
We handle the education end to end. Building it, running it, marking the work and issuing the qualification. Nobody at your end writes a curriculum, records a video or chases completions.
The basics that apply to any AI tool: briefing a job so it lands, spotting an answer that is confidently wrong, and keeping the bill tracking the work. The core is the same for every business, which is what makes it the cheapest part, and it works with whatever your team already has open rather than only with NeuralVault.
Per seat.Built on your own systems and your own work: getting your methods into the tools, writing the instruction file, recording a job so the steps are captured, and running NeuralVault properly. This is the part large firms are spending real money on, and it costs more, which is why most businesses put a handful of people through it rather than everybody.
Per seat or per department. Costs more.We go through what you actually run before writing anything, so each department is taught on its own stack instead of a generic one. It is charged separately because it is real work rather than a sales call, and it is the difference between a course about AI and a course about your business.
Charged separately.One thing we will not overstate: we are not an accrediting body and will not pretend to be one, so this is not a regulated qualification. What a person leaves with is evidence of the work they did, in a form another employer can read, which is more than a certificate of attendance has ever been worth. None of it is on sale yet, there is no date, and there is no figure on this page because there is not one we would stand behind. If it matters to you, tell us and we will raise it at a consultation. What clients ask for decides what gets finished first.