Your first AI workshop: less watching, more doing

Choose one outcome for the first session

A first AI workshop should help people complete and check a small, useful task with an approved tool. Choose a task the audience understands, prepare appropriate material and leave enough time for everyone to attempt it. Do not make the first session a tour of every possible application.

The agenda below is a proposed two-hour format for a small workplace group. Adapt it to learner experience, accessibility requirements and the amount of feedback available. It is a teaching plan, not evidence that two hours establishes workplace competence.

Prepare the conditions for practice

Before the workshop, confirm the tool, account access and information rules with the organisation. Ask learners to complete a short access check. Prepare synthetic or approved examples and a fallback activity if someone cannot use the environment.

Define the acceptable output and the review checklist. For an illustrative internal summary task, the criteria might include factual consistency with the source, clear separation of facts and suggestions, and visible unresolved questions.

Give managers a short note explaining the outcome and the follow-up. If learners are expected to practise at work afterwards, someone needs to make that time and support available.

A two-hour agenda you can adapt

0–15 minutes: establish the task and boundaries. Explain what the group will produce, what information may be used and who remains responsible for checking the output. Invite questions about the task before opening the tool.

15–30 minutes: demonstrate a complete attempt. Show the brief, the draft and the review. Include a plausible error or omission and explain how you identify it. Avoid presenting a polished final answer as though it arrived without any checking.

30–55 minutes: individual practice. Give learners a fresh but comparable example. Ask them to write the instruction, inspect the response and mark changes before revising it. Circulate to understand where people are getting stuck.

55–65 minutes: pause and compare. Discuss what varied across outputs. Take a short break if it helps the group reset. The aim is to notice differences in the work, not to rank learners by how impressive their first prompt sounded.

65–90 minutes: review and improve. Use pairs or facilitated discussion to apply the same checklist. Ask people to explain one factual check and one judgement they made. Then let them improve their output.

90–110 minutes: apply the learning to a suitable task. Have participants identify a real workflow they could try later within the approved boundaries. If no suitable task is available, record that as a planning issue rather than forcing an example.

110–120 minutes: agree the next practice. Record what each person will attempt, what support they need and when the result will be reviewed.

What to take away

Provide the exercise, an acceptable example, the review checklist and the agreed support route. Keep the AI prompt template attached to its checking steps. A prompt copied without context can lose the parts that made it useful.

Do not claim that attendance alone demonstrates independent capability. If you need evidence, ask learners to complete a suitable fresh task and explain their checks. Record the limits of what was observed.

Make the second conversation part of the design

Arrange a follow-up to look at examples and barriers. Ask where the workflow helped, where it added effort and where people chose not to use it. Use those answers to decide whether the next activity should be more practice, a process decision or a different task.

The 30-day adoption plan extends this workshop into a pilot. Explore AI foundations for every role or contact Experrt to shape a session around your team’s starting point.