A well-run AI workshop can be useful. It can remove anxiety, show people what the tools can do and give a small team a common vocabulary. It cannot, by itself, create a working capability.
The gap becomes clear a few weeks later. One person is using a tool every day. Another is unsure what information can be shared. A third has decided the results are unreliable after one poor response. The team attended the same session, but they are now learning in separate directions.
Capability emerges when people use a tool on real work, compare judgements, see where it fails and develop shared standards. That is a management task as much as a training task.
Start with a work scenario
For a 12-person team, the best training material is rarely a generic list of prompts. Use a familiar job: preparing a client update, organising a proposal, summarising a meeting, reviewing a long policy document, drafting a standard response or preparing a training outline.
Choose work that is real but safe to use in a learning setting. Remove confidential details where necessary. Ask participants to compare the old method and the AI-assisted method, then discuss where the output needed correction. This makes the conversation about work quality, not entertainment.
Teach checking, not just prompting
Prompting gets attention because it is visible. Verification is where professional capability begins.
Participants should learn to ask: What is this answer based on? What could it be missing? Does the tone suit the audience? Has it invented a fact? Is the reasoning plausible? Would I be comfortable attaching my name to this?
These are not technical questions. They are the same judgement habits good professionals already use when reviewing a junior colleague’s draft or checking a spreadsheet prepared in a hurry. AI changes the speed and volume of output. It does not remove the need for judgement.
Give the team a small common language
Teams benefit from a few simple phrases they can use without embarrassment. “This is a first draft.” “I need a source for that.” “This contains client information, so stop.” “This is a decision for a person, not a tool.”
Such phrases make it easier to raise a concern early. Without them, the person who feels uneasy may stay silent because they assume everyone else understands the technology better.
The aim is not to make every employee an AI specialist. It is to create a shared floor of competence: people know what the tool is for, what it is not for and when to ask for help.
Learn in short cycles
Rather than a single half-day event, run a short cycle over four weeks. In week one, introduce the chosen use case and the information boundary. In week two, ask people to use the tool on one defined task. In week three, review examples that worked and examples that did not. In week four, agree on a simple standard for the workflow and decide whether to continue.
This sequence creates feedback from the actual workplace. It also prevents the training from becoming a performance where people nod along, then return to the same habits on Monday.
Make one person accountable for the practice
The facilitator does not need to be a technical expert. They need to keep the learning connected to the work. They collect questions, make sure the team’s rules are visible and ask whether the promised benefit is materialising.
Where an outside trainer is involved, their role should be to help the team establish this practice, not to become the permanent owner of it. A good programme leaves behind a team that can keep improving its own use.
Let competence show itself in the work
The evidence of useful training is not a certificate, a group photo or a list of tools covered. It is a better routine: more consistent first drafts, fewer missed actions after meetings, clearer client communication, or a team that knows when not to proceed.
This is why AI capability should be considered part of organisational readiness. It is built through repeated, situated practice. For a fuller view of the conditions around it, see AI readiness versus digital transformation maturity. A workshop can open the door. The work after the workshop is what makes the capability real.