Before the AI workshop, prepare the work

Choose a task people actually recognise

“Learn AI” is a broad ambition. “Prepare a first draft of our monthly activity report from approved notes” gives a workshop something concrete to work with. Start by identifying two or three recurring tasks that people understand well enough to judge the result.

Good starting exercises have a clear input, a useful output and a person who can check whether the work is right. A short document summary, a comparison of two approved versions or an outline for a presentation is easier to review than an open-ended request to improve an entire department.

• What information does the task start with?
• Who uses the result?
• What must the result include?
• Which errors would make it unusable?

Bring representative material, with the sensitive parts removed

A familiar example helps participants understand the method. It does not need to contain live customer, employee or financial records. Prepare a shortened, anonymised document or a synthetic dataset that preserves the structure of the task.

Remove more than names. Account numbers, contact details, unusual combinations of facts, document properties and hidden spreadsheet sheets can also reveal information. Use the organisation’s approved process and tools for preparing training material. If that approval is not clear, use a purpose-built example instead.

Check access before the session

Confirm the selected AI tools with the trainer before asking everyone to create accounts. Account type, feature availability, file limits and organisational restrictions can affect an exercise. The plan should match what participants are actually permitted to use.

Test sign-in, the internet connection and any required file access on the devices that will be used. For an online workshop, check the meeting platform and whether participants can comfortably switch between the demonstration and their own work.

Understand the group’s starting point

A team can share a job title and still have very different levels of confidence. A short pre-session check is more useful than a broad label such as beginner or advanced. Ask whether participants have used an AI assistant, what they tried and where they became uncertain.

Share language preferences and accessibility needs early. The workshop can then use a sensible pace, familiar terminology and a balance of demonstration and practice that suits the group.

Define success as something observable

A good end-of-session task might ask participants to write a brief, produce a draft and explain the checks they applied. This shows whether they can use the method, not merely whether they followed the demonstration.

Avoid promising a productivity percentage before there is a baseline. If the team wants to measure improvement later, choose a comparable task, record the existing process and agree how quality will be assessed alongside time. Faster work is only useful when the result is still fit for its purpose.

Plan the next small application

Before the workshop ends, identify one suitable task that participants can revisit at work. Assign a reviewer, retain the approved prompt and record the changes needed for the organisation’s context.

A useful follow-up conversation asks what worked, what needed correction and which information was missing. Those observations can guide the next training session more reliably than adding another long list of tools.