A team trained together builds the same checks and starts catching each other's mistakes. Choosing the cohort shape, mixing routes, the group's funding ceiling and its time.
Running a Data and AI Apprenticeship Cohort for Your Team
By James Cotton · Last updated · 5 min read
Part of our topic guides on AI Skills for Business and Government-Funded Data & AI Training.
By James Cotton, Founder, iO-Sphere
How a cohort works, and why together beats one at a time
A cohort is a group of apprentices learning in the same sessions. Those sessions run on professional discussion, peer challenge and networking, not self-study. Our coaches are practitioners in data and AI, and the group does most of its learning by arguing through real problems in front of each other.
It sounds like this. An analyst explains why she dropped a month of figures from a trend. A colleague asks whether the gap is real or a pipeline fault. The analyst leaves with a sharper check, the colleague with a question to put to her own work, and everyone else in the session heard the exchange.
Over a programme, a team converges. They settle on the same checks for what counts as a trustworthy number or a usable model output, then apply them to each other in ordinary meetings, long after a coach last saw the work. The same people sent on separate courses each come back improved, alone.
Employers already favour training the people they have. The Office for National Statistics' 2026 survey of AI in UK businesses found it the most common way businesses bring AI skills in. Among businesses that cite a lack of AI expertise as a barrier, around 62% train or retrain existing staff; among businesses reporting no barriers, the figure is around 26%.
Open or closed: choosing the shape
An open cohort is one where learners from different companies and industries train together. An SME with one or two people joins an open cohort: there is no minimum on their side. The mix is the product. Your analyst argues through problems with people who do the same job elsewhere, under different pressures and with different data.
What comes back is practice your own team could not have grown alone. That breadth suits a small group, or a larger team whose manager wants them tested against the wider field. Nobody in the session shares their assumptions by default.
A closed cohort is built solely around one organisation, and for one we look for 10 to 12 people. The group works on the organisation's own cases, so the discussion, the challenge and the habits all fit your business. We build these as custom programmes. The trade is the outside view: everyone in the room does things the way you do.
Each shape is a real choice. The open cohort is not the economy version of the closed one: it teaches through breadth, the closed cohort through depth. What decides it is whether you want your people learning mainly from colleagues who share their context or from peers who do not.
Mixing roles and routes in one group
A cohort does not need everyone on one route. A mixed team can take different apprenticeships and still learn side by side: analysts on Advanced Data & AI, people building pipelines on Data Engineering, and a manager on Data & AI Governance or Data & AI Strategy. The shared sessions are where their different work meets.
Funding sets no cap on numbers. No rule in the 2026-27 funding rules limits how many apprentices one employer may have, so a team of ten can all be apprentices at once. Each route, its level and who it suits are laid out in our guide to AI apprenticeships for employers, and our apprenticeships page lists every one we run.
The money: ceilings for a group
A funding band is the most government funding will contribute for one learner on a standard, and a group's ceiling is that band times the headcount. On the bands Skills England published on 24 September 2026, ten people on Advanced Data & AI give a ceiling of £150,000, and ten on Data & AI Essentials £130,000. Neither figure is a quote. The other routes' bands are in the routes guide linked above.
Size is a negotiating point. Paragraph 210.1 of the 2026-27 apprenticeship funding rules names cohort size as a legitimate basis for negotiating price, with any saving coming from within the band.
Age and timing both go into the plan. When we plan a cohort we list each person's role, route and age, since a young starter is funded on different terms from a colleague over 25. The plan also sets the start date against your levy year.
The time to protect
Each apprentice needs protected off-the-job training time: 370 hours on Advanced Data & AI, Data & AI Governance and Data & AI Strategy, and 348 hours on AI Transformation, the minimums Skills England publishes for each standard. Across a closed cohort of ten on Advanced Data & AI, that is 3,700 hours of working time the business commits before the first session. Plan it early.
Scheduling is a team decision. Agree with each line manager when their people step away and who covers the work while they do. If four people in one department are away at once, that department feels it, and the plan should say how it copes.
Managers have a load too. Each one needs to protect the hours and stay close to the work the apprentice produces. A manager with three apprentices in the cohort does that three times over, so ask them before you commit; a cohort that loses its protected time loses most of what made it worth running.
Starting a cohort
We ask for three things. First, the roles, with the data work each person does now. Second, a named sponsor who owns the cohort and can hold the time. Third, the shape you want, open or closed. From there we match each person to a route and plan the group with you.
The sponsor's first task comes before any session. They decide what the team's work should look like differently by the end of the programme, and write down how they will see the change in the business, so the cohort is judged on the team's work; our guide to measuring workplace training outcomes covers choosing those measures.
Frequently asked questions
Can people in the cohort start on different dates?
Training side by side is what builds the shared checks, so we plan the group around learning together. If a start date does not suit everyone, raise it when we scope the cohort, before anything is booked.
We do not pay the levy. When should we reserve funds?
Close to the start. Under the 2026-27 funding rules, reservations of funds apply only to non-levy employers, and a reservation lapses after three months if it is not used, so reserving too early for a cohort whose start then slips means reserving again.
Can we spend our levy on staff from a subsidiary?
The route the funding rules give for spending levy on another employer's apprentices is a levy transfer, with the annual limit and the steps on that gov.uk page. Whether a subsidiary counts as a separate employer for that purpose depends on how your group's levy account is set up, so we check that with you before planning places.
Will it fit your cohort?
Tell us your cohort size and what your people need to be doing by next year, and we will say whether we are the right shape, including when we are not.