The 4-Phase AI Literacy Playbook for HR (Why 93% of Rollouts Fail)

By Johannes Sundlo, AI & Future of Work Advisor. I help leaders and HR teams turn AI into adoption that sticks, through keynotes, workshops and change programs.

Here is the number that should stop every HR and leadership team cold: organisations spend roughly 93% of their AI budget on data, technology and infrastructure, and just 7% on the people who have to use it. Then they wonder why adoption stalls.

Most AI rollouts do not fail because the technology is bad. They fail because they were run as a tech project, not a change project. A licence got bought, an email went out, a one-off training happened, and three months later usage flatlined. AI literacy is not a webinar. It is a structured rollout that earns adoption phase by phase.

This is the 4-phase playbook I use with organisations. It takes three to four months done properly, and it is built so capability scales from the top down without becoming a bottleneck.

Phase 1, Leadership alignment & defining success

Before anyone is trained, leadership gets hands on the tools themselves and agrees on two things: why AI matters here, and what success looks like in numbers. Skip this and every later phase drifts, because no one can say whether it is working.

• Executives use the tools directly, no delegating the understanding.
• Agree a shared narrative for why this matters to your business.
• Define success metrics up front (time saved, quality, adoption rate).

Phase 2, Train the trainers

You cannot train a whole organisation from the centre, and you should not try. Select 10–15 internal AI ambassadors from across functions and give them intensive, multi-week training. They become the people who scale instruction, and, more importantly, the trusted local face of it.

• Pick ambassadors who represent different parts of the org, not just IT.
• Invest deeply in this small group, they multiply your reach.
• They carry context the central team never could.

Phase 3, Manager enablement

Managers make or break adoption. If a manager is anxious about risk, the team will not touch the tools; if a manager models use, the team follows. This phase gets managers using AI themselves and gives them simple principles for guiding their team without blocking experimentation.

• Managers learn by doing, not by reading a policy.
• Equip them to encourage use and hold sensible boundaries.
• The goal is permission with guardrails, not fear.

Phase 4, Employee training & real use cases

Now the broader organisation gets trained, ideally by those internal ambassadors, not an outside voice, and every team commits to one or two concrete, documented use cases. Documentation matters: it turns a one-off “that was cool” into a repeatable habit other teams can copy.

• Train through internal ambassadors for trust and context.
• Each team ships one or two real, documented use cases.
• Use cases become the proof and the playbook for the next wave.

Why the order matters

The phases are sequential on purpose. Train employees before leadership has defined success, and you get enthusiasm with no direction. Skip managers, and adoption dies at the team level. The companies that get this right, the Modernas and Spotifys of the world, treat AI literacy as change management with a technology component, not the other way round.

And this is foundational literacy, not the finish line. It is the floor that everything else, agents, deeper integration, new ways of working, gets built on. If the vocabulary in here is shaky for any group, start them with the AI dictionary for HR. For the leadership-specific view, see the AI guide for leaders.