Only two apprenticeship standards name AI in the title. Here is every AI-facing route, which ones we deliver, and how to choose by the capability that outlasts the tools.
AI Apprenticeships for Employers: Every Route Compared
By James Cotton · Last updated · 6 min read
Part of our topic guide on AI Skills for Business.
By James Cotton, Founder, iO-Sphere
Every AI-facing standard, in one table
Demand has grown fast. Department for Education figures published on 16 July 2026 show starts on the Digital route more than doubled, from 14,760 in 2020/21 to 31,410 in 2024/25. The standards that build AI capability are harder to spot, because few of them say so on the cover, so the table lists every one with an AI claim, ours and those we do not deliver.
| Standard and code (Skills England, 24 September 2026) | Level | Funding band | Typical training period | "AI" in the title | Our programme on it |
|---|---|---|---|---|---|
| Data technician, ST0795 | 3 | £13,000 | 24 months | No (AI is in its knowledge) | Delivered by iO-Sphere as Data & AI Essentials |
| Business analyst, ST0117 | 4 | £18,000 | 18 months | No | Delivered by iO-Sphere as AI Transformation |
| Data analyst, ST0118 | 4 | £15,000 | 24 months | No | Delivered by iO-Sphere as Advanced Data & AI |
| Data protection and information governance practitioner, ST0967 | 4 | £10,000 | 18 months | No | Delivered by iO-Sphere as Data & AI Governance and Data & AI Strategy, both on ST0967 |
| Artificial intelligence (AI) and automation practitioner, ST1512 | 4 | £18,000 | 18 months | Yes | Not delivered by iO-Sphere |
| Data engineer, ST1386 | 5 | £19,000 | 24 months | No (AI is in its knowledge) | Delivered by iO-Sphere as Data Engineering |
| Machine learning engineer, ST1398 | 6 | £22,000 | 24 months | No | Not delivered by iO-Sphere |
| Data scientist (integrated degree), ST0585 | 6 | £19,000 | 36 months | No | Not delivered by iO-Sphere |
| Artificial intelligence (AI) data specialist, ST0763 | 7 | £17,000 | 24 months | Yes | Not delivered by iO-Sphere |
For the five we deliver, the register's typical periods range from 18 to 24 months up to the final assessment, against our own programmes of around 15 months' training followed by three months of assessment.
What the duties say, whatever the title
Titles mislead in both directions. The Level 3 data technician standard has no AI in its name, yet its version 1.1, in force from 20 August 2025, asks the apprentice to know "the relationships between data, machine learning, Internet of Things, Artificial Intelligence and Generative AI" and "the ethical use of data ... with Artificial Intelligence". We deliver it as Data & AI Essentials, and the Level 3 guide goes further into that route.
The Level 5 data engineer standard works the same way. AI appears in its knowledge only. Its current version, 1.0, accepts starts until 17 December 2026, and we deliver it as Data Engineering.
The business analyst and data protection standards, both at Level 4, name AI nowhere in their title or duties. Skills England summarises the first as "working with organisations to improve their information systems" and describes the second as providing "regulatory and technical advice providing assurance to key stakeholders and regulators". Neither depends on which model a team uses.
We deliver the business analyst standard as AI Transformation. The data protection standard, whose version 2.0 took effect on 9 September 2026, is the base for two of our programmes, Data & AI Governance and Data & AI Strategy. The AI governance in them is how we teach it; the standard's own text does not use the phrase, and the data protection apprenticeship guide covers the mechanics.
The Level 4 AI and automation practitioner standard names AI in its title and throughout, with duties built around the "safe and responsible use of automation, integration, and AI tools". Version 2.1 applies to starts from 22 May 2026, and the standard was first approved on 10 December 2025. It is the right choice when putting those tools to work is the role itself. We do not deliver it, so an employer filling that role should find a provider who does.
For someone in finance, operations, marketing or compliance, with AI as one tool among several, the business analyst or data protection standard fits better. The Level 4 guide compares the Level 4 options in more depth.
Level 6 and 7, and when to look past this page
Above Level 5 there are three more standards with an AI claim, two at Level 6 and one at Level 7, and iO-Sphere delivers none of them. Under the 2026-27 funding rules, the Level 7 AI data specialist standard has been funded since 1 January 2026 only for starts aged 16 to 21, or 22 to 24 with an EHC plan or care-leaver status, which rules it out for most experienced colleagues; for degree-level routes, the graduate programmes guide is the better start.
Then fit the person
Hiring is the rarer path. A Department for Science, Innovation and Technology survey fielded in 2024 and published in January 2026 found that 11% of employers had trained staff on AI in the previous twelve months, while 4% had tried to recruit someone to work with AI in the previous three years. The Office for National Statistics reported in July 2026 that recruiting AI-skilled staff ran from about 2% of the smallest businesses to about 10% of those with 250 or more employees.
A colleague already doing the work brings the context an AI output has to be checked against. They know which numbers look wrong. A new hire can take the same route, whether a young apprentice or someone further into a career, provided the role involves the work the standard describes.
Several people with the same need can start together as a cohort. Whether a start draws on the levy or not, the standard that fits stays the same.
Choose by what stays after the tool changes
An apprentice who starts this autumn will reach the final assessment after the tools on today's brochures have changed. Three capabilities will still matter.
Framing the problem. Before anyone opens a tool, someone has to decide what the business is asking and what a useful answer would look like. A programme builds that only when the apprentice sets the question on real work.
Checking the output. An AI answer arrives fluent and sometimes wrong. The habit to build is testing it against the source data, a second method or a colleague who knows the ground.
Explaining the result. A finding a manager cannot follow changes nothing. The apprentice needs practice presenting to the people who will act on it.
That leaves one question to put to any provider, including us. How does the programme assess real work for each of the three, and who outside the training sees the result?
Frequently asked questions
How do we start one?
Begin with the person: someone whose role already involves data work, or a hire into a role that does. The provider then runs an initial assessment of what that person knows, and a training plan is agreed with you and the apprentice before a start date is set. The apprenticeships page lists the programmes to start from.
Can an existing analyst do the AI Transformation route?
Often, yes. AI Transformation is built on the business analyst standard, so it suits an analyst whose next job is improving how the organisation's information systems serve it. If the work ahead is deeper analysis of the data itself, Advanced Data & AI is the programme to compare, and the data analyst employer guide sets out that route. The deciding question is which of those two jobs the analyst is moving into.
Do we need a technical manager?
No. The standards behind AI Transformation, Data & AI Governance and Data & AI Strategy name no AI in their duties, so what the apprentice brings back is business and governance work. What helps most is a manager who can hand over a real problem and ask how the answer was checked.
Build a data-literate workforce
Data apprenticeships funded through the Growth & Skills Levy: £0 for any apprentice under 25, and most of the cost covered for older starts.