AI in HR is no longer a consideration for the future. Most UK organisations now have employees using AI tools in some form. Far fewer have a structured way to help them use those tools well. Research from the Department for Science, Innovation and Technology found that only 21% of UK workers feel confident using AI at work, while just one in six UK businesses are actively using it in any meaningful way. The gap is not one of access. It is one of capability.

This shows up clearly at the organisational level too. McKinsey research has found that 90% of companies report investing in AI, yet fewer than 40% report any meaningful impact on their bottom line. The tools are there. The redesigned workflows and the skills to use them properly, more often than not, are not.

For L&D teams, this puts AI capability building in an unusual position: not a nice-to-have alongside other training priorities, but the difference between AI adoption that actually changes how work gets done and AI adoption that stalls at the surface. That starts with:

• What AI capability actually means, and why it looks different depending on role
• Where to start if there is no formal programme yet
• How to design training that changes behaviour rather than just raising awareness
• How to make that training repeatable, so it scales without burning out the L&D team
• What a mature, well-run programme looks like in practice

Explore: AI and the Future of Work: What’s Actually Happening to UK Jobs

For most organisations, “AI training” means one session: a lunch-and-learn, a webinar, a module bolted onto onboarding. CIPD’s Autumn 2025 Labour Market Outlook found that 61% of employers now allow generative AI use at work, but only 35% provide any training, and just 31% have worked on a formal AI usage policy in the past year. Staff get the tools without the structure to use them well.

CIPD research from early 2026, surveying over 1,300 HR professionals and senior decision-makers, found many organisations are adopting AI faster than they’re building the capability to support it. Skills development ends up reactive, disconnected from how AI is actually being used.

Even where training exists, it often doesn’t stick. A McKinsey study of Microsoft 365 Copilot adoption found nine in 10 employees agreed formal training would help, yet seven in 10 ignored onboarding videos entirely, learning instead through trial and error or from colleagues. Knowing training matters and using it are not the same thing.

The fix isn’t more sessions. It’s repetition, relevance to real tasks, and clarity on what different roles actually need to know how to do.

“AI capability” gets used loosely, often to mean little more than knowing how to write a prompt. In practice, it covers three distinct layers:

• Tool-level skill – using specific AI tools competently for a given task
• Judgement – knowing when to trust AI output, when to verify it, and when to override it
• Governance awareness – understanding data privacy boundaries and appropriate use within the organisation

Most training only addresses the first. The other two are what separate confident, safe use from misuse.

These needs also differ sharply by role. PwC’s 2026 Global AI Jobs Barometer found that the most AI-exposed junior roles are now seven times more likely than the least-exposed roles to require senior-level skills such as leadership and judgement. Work that once served as a junior apprenticeship, learned by doing the routine parts first, is increasingly automated, which pushes judgement-based skills earlier into a career than training programmes typically account for.

This is why a single AI training module rarely fits everyone. Leadership needs strategic and risk oversight. Managers need to know how to support and evaluate their team’s use of AI, and spot misuse early. Individual contributors need practical tool skill paired with the judgement to apply it well in their specific role. As McKinsey puts it, reskilling for AI isn’t about turning everyone into an AI engineer. It’s building fluency alongside the human skills, like critical thinking and quality assurance, that complement it.

Most organisations are earlier in this journey than they think. No formal programme, ad hoc tool use across teams, no clear policy on what’s acceptable. That is a normal starting point, not a sign of falling behind. World Economic Forum data shows 50% of the global workforce has already completed some form of training, reskilling or upskilling, up from 41% in 2023. Structured capability building is still in its early stages almost everywhere.

There is also a clear national push to build on. The UK government’s AI Skills Boost programme commits to free, practical AI training for 10 million workers by 2030, underpinned by Skills England’s new AI foundation skills benchmark. Organisations starting now have a reference point for what “good” looks like, rather than having to define it from scratch.

Practically, the first steps are small and deliberate:

• Run a baseline skills audit – understand what people are already doing with AI, formally or informally, before designing anything
• Identify a handful of quick-win use cases – specific tasks where AI can save time now, rather than a broad rollout across every function at once
• Set minimal governance guardrails early – clear rules on data handling and appropriate use, even if basic, before scaling any training programme

The common mistake is waiting for a fully-formed programme before starting. It is better to begin small, learn from what works, and build from there.

Once an organisation moves past ad hoc use, the design of the training becomes the deciding factor between a programme that sticks and one that gets ignored.

A few principles separate the two. As AI trainer and consultant Ashley Couto has noted, training works best when it’s hands-on and tied to real job tasks, not generic content. Practice matters more than exposure: showing someone a tool once is not the same as having them use it on their own work with feedback. And judgement needs to be taught alongside mechanics, since knowing how to use a tool and knowing when to trust its output are different skills.

Reinforcement over time matters too. LinkedIn Learning’s Workplace Learning Report found that “career development champions,” meaning organisations with the most mature, effective career development programmes, are 32% more likely to be running AI training programmes this year than other organisations. Structured development and sustained AI training tend to go together.

Designing one good training session is achievable. Designing one for every team, every quarter, without burning out the L&D function, is a different problem entirely.

Deloitte’s 2026 Global Human Capital Trends research found that 85% of leaders say their organisation’s ability to adapt at speed is critical, yet only 7% believe they’re actually leading on that front. Most organisations know they need to move faster than they currently can.

Bespoke training built fresh for each team closes that gap slowly, if at all. It does not scale, and it puts unsustainable pressure on whoever is building it. A playbook approach solves this differently: a core structure, covering the fundamentals and governance basics that stay constant, with room built in for teams to adapt the application to their own workflows. Consistency where it matters, flexibility where it doesn’t.

Organisations that get AI capability building right see it show up in performance, not just in training completion rates.

PwC’s 2026 AI Jobs Barometer found productivity growth is 40% higher at the most AI-exposed companies compared with the least. PwC’s UK-specific analysis found the UK wage premium for AI-skilled workers tripled in a single year, from 11% in 2024 to 34.2% in 2025, as demand for people who can use AI well continues to outpace supply.

These are organisation-level outcomes, built from the kind of structured, repeatable approach covered above rather than a single training push. What that looks like in practice, for one organisation working through exactly this process, is covered next.

At Avado, we help L&D teams move beyond one-off AI awareness sessions to build capability that actually changes how people work. Our AI for HR Bootcamps give teams a structured, practical starting point, whether an organisation is beginning its AI upskilling journey or scaling a programme across the business.

Explore our AI for HR Bootcamps and start building AI capability that sticks!