Most employers don’t come to us asking about the Level 4 AI and Automation apprenticeship. They come asking how to get their team using AI properly, and the apprenticeship turns out to be the answer. The ones who walk in naming it have usually done a lot of reading first.

So, plainly. An AI apprenticeship is a government-funded, work-based training programme where someone already in your business, or someone you’re bringing in, spends a minimum of six hours a week learning and the rest of the week doing the job. With us, the AI and Automation Level 4 runs about twelve months of training, followed by a couple of months for the end-point assessment at the end, where the training has finished but the assessment is still running. It teaches people to design, build and maintain automated workflows and AI agents in a real business. No coding background needed.

Skills England ran the project that created that standard, and I led the R&D on it from the industry side. That meant sitting down with dozens of high-growth SME business leaders and asking about their barriers, their blockers and their ambitions. Not just the technology, and not just the skills. We talked about business outcomes and the other pressures leaders are carrying, because you can’t design a training programme that changes how a business works if you only understand one corner of it.

Two things came out of that work more strongly than anything else, and they shaped the whole standard.

The first is that businesses want outputs, not certifications. Nobody wanted to wait twelve months for a qualification and then start seeing value. They wanted value back in the business early and then compounding. So the programme is built to deliver something usable in the first few months and keep building on it.

The second took longer to get to.

This is a generalist qualification, and that’s deliberate

Look at what’s on the market and you’ll find AI for marketers, AI for HR, AI for policy teams, AI for finance. AI as a specialism you bolt onto a function.

We do that too. Our other apprenticeships and courses have AI running through them, and they’re specialist: data, software development, product. If you need depth in a function, that’s the route.

This one is different on purpose. Creating and maintaining automated loops and agents is becoming part of everyone’s job, in every function, and the thing that makes someone good at it is the domain knowledge they already have. Your operations manager knows where the process actually breaks. Your finance lead knows which reconciliation eats three days a month. That knowledge is the valuable half and you already have it in the building.

What people don’t have is the other half. Enough understanding of data to know when the output is rubbish. Enough about privacy, security and ethics to not create a problem. Problem framing, testing, business analysis, product thinking, project management, working in small releases rather than big launches, and handling the change that lands on the team around them.

Those are general skills. Good practice and rules that apply to everything, regardless of what the person does all day. That’s why this one is for everybody, and I mean that literally rather than as a marketing line. Marketing, operations, delivery, finance, policy. Same skills, different domain knowledge on top.

The skills getting neglected while everyone looks at the models

There’s a habit of treating AI adoption as an automation exercise, where the whole goal is doing the existing job with fewer hands on it. That’s a small piece of it.

The businesses getting real value are building genuine capability around the AI work. Someone who can test properly. Someone who can do business analysis and work out whether a thing is worth building before it gets built. Someone who can manage a product rather than a project. Those skills are quietly going missing while the attention sits on the tools.

That’s why the AI Product Manager role matters, and why we’re putting more into it. It’s also why the supporting apprenticeships around the AI one are worth looking at, rather than treating the Level 4 as the whole answer.

Creating and maintaining automated loops and agents is becoming part of everyone’s job, in every function.

Where the Level 4 sits, and what the other pathways are for

The digital and AI apprenticeship pathways cover a reasonable spread, and a lot of businesses reach for the wrong one.

Data and analytics foundation (Level 3 and Level 4)

Data Technician Level 3, Data Analyst Level 4. Roles like data analyst, junior BI analyst, automation assistant. This is the route out of spreadsheets: Python, SQL, GDPR compliance, basic machine learning, automation tooling to speed up reporting. Start here if the business problem is that nobody can get a straight answer out of your own data.

AI and workflow automation (Level 4)

The AI & Automation Level 4 apprenticeship, standard ST1512. Roles like AI champion, workflow automation developer, junior prompt engineer. Workflow design, no-code automation platforms, risk assessment, governance. For most businesses, most of the time, this is the right one, because it’s about the doing: using these tools productively, well, and with some discipline about how.

Engineering and software development (Level 4 and Level 5)

AI engineer apprenticeship, Software Developer Level 4, Data Engineer Level 5. Git, APIs, databases, cloud infrastructure, ETL pipelines, secure coding. This is for businesses building custom software, not businesses adopting AI into how they work. Worth being clear-eyed about which of those you are.

Degree-level (Level 6 and 7)

AI degree apprenticeships exist and they’re interesting. If you have a real need for that depth you’re probably a tech business that’s already doing well, and you’ll know it. For most of us it’s academic, and the Level 4 is where the need actually sits. I’d rather say that than sell you something bigger.

On providers, several operations deliver AI apprenticeships in the UK, including us and larger ones such as Multiverse. Worth talking to more than one.

What it costs

If your payroll is under £3 million, you don’t pay the levy and you can stop worrying about that part. Anyone who does pay into it already knows they do.

For everyone under that threshold, from 1 August 2026 the government funds 100% of training and assessment costs for any apprentice aged 16 to 24 at the start of their training. Nothing to pay for the training at all.

For apprentices aged 25 and over, the government funds 95% and you pay the remaining 5%. The funding band for the AI and Automation Level 4 is £18,000, so your share is no more than £900 across the whole programme, and in practice less, because the 5% is calculated on the agreed price rather than the band maximum. Spread across the programme that’s under £70 a month.

Either way you pay their wages, and that’s the real number to plan for.

If you are a levy payer, your levy funds cover the full training and assessment cost up to the band maximum. Two changes worth knowing about: from August 2026, if you exhaust your levy pot the co-investment rate is 25% rather than the 5% it used to be, and unspent levy now expires after twelve months instead of twenty-four. Sitting on it has got more expensive.

What the time commitment actually looks like

The figure still quoted everywhere is 80/20, and it hasn’t been true since 2022. The rule is now a minimum average of six hours a week of off-the-job training for a full-time apprentice, meaning someone working 30 hours a week or more. It’s a flat number rather than a percentage, so a person on a forty hour week is doing under 20% of it. Part-time apprentices do fewer hours pro-rata over a longer programme.

More useful than the number is what the six hours consists of.

Across a month there’s group training, group coaching and one-to-one coaching. Around that, the majority of the learning time is the apprentice working on actual AI and automation projects inside your business. The learning is aimed at the work, so what they build is work you wanted doing.

It isn’t six hours a week at college. Done with us, it’s six hours a week of supported improvement to your business while they’re being trained doing it.

Who should do it, and who should be leading it

Two patterns come up often enough to be worth naming.

The first is a leader who comes in intending to do the apprenticeship themselves, and works out partway through the conversation that it’s better if two of their people do it and they lead what those people build. That’s usually the right call. The apprenticeship suits someone doing the doing.

The second is the reverse. A leader who came in expecting to hand it to someone else, then wants it themselves once they can see how the role changes.

Both are fine. What doesn’t work is a leader who commissions the work and then can’t tell whether what comes back is any good. If you’re going to lead people building automations and agents, you need enough understanding to direct it, and you need it faster than a year. That’s what our AI leadership training is for: a short course that gets leaders to the point where they can get more out of the people doing the apprenticeship, which pulls the return forward.

The other thing worth saying is that you don’t always have to do the whole thing. Where someone already has real AI expertise, prior learning can be recognised and the programme can run in as little as eight months.

Why building it in-house works

Hiring experienced AI people externally is expensive and slow, and you’re competing with businesses that will outbid you.

Training existing staff gets you something recruitment can’t: people who already know your business, your customers and where your processes leak, now equipped to build on that. Bringing new apprentices in alongside them gives you the other half, people learning to build loops and agents from the ground up who’ll pick up the domain knowledge from the people around them. The two directions feed each other, and you end up with career paths inside the business rather than a hiring problem every eighteen months.

Common questions

What is an AI and automation apprenticeship?

A government-funded, work-based programme that qualifies someone as an AI and Automation Practitioner against Skills England standard ST1512. With us it runs around twelve months of training plus a couple of months for end-point assessment. Learners apply AI tools, automate workflows and govern AI use responsibly through real projects in their actual job. No coding background needed.

Can an SME get funded AI training?

Yes. If your payroll is under £3 million you don’t pay the levy. Training and assessment is fully funded by government for apprentices aged 16 to 24, and 95% funded for those aged 25 and over, leaving you no more than £900 across the programme.

Does the apprentice need a technical background?

No. The standard is written for non-technical professionals in operational roles. The skills that matter most are the domain knowledge they already have.

Can existing employees do it, or is it only for new hires?

Either. Funding covers existing staff as well as new recruits, and for most businesses training someone already in the team is the faster route to value.

How much time does it take out of the working week?

A minimum average of six hours a week for full-time apprentices, pro-rated for part-time. With us, most of that time is spent on real projects in your business rather than in a classroom.