If every successful output in a workshop comes from the trainer’s screen, you have seen a demonstration, not evidence of team capability. A useful agenda lets participants attempt the work, receive feedback and try again with a new example. The sample below shows what that can look like for one narrow office task.
Research reviewed: 1 October 2026. The worksheets and scenarios below are EverX editorial tools, not validated assessment instruments or reported client results.
Choose one outcome before writing the agenda
Example outcome: participants can turn fictional meeting notes into an action list while preserving unconfirmed owners and dates. They must explain which items need clarification before the list is circulated. This is deliberately narrower than ‘master AI for productivity’. A three-hour session cannot credibly promise mastery across tools, governance, automation and every job role.
Confirm the audience’s starting level, language, tool access and approved information before the day. Where someone cannot access the tool, prepare a paired activity or a printed draft-review exercise. Do not let setup consume the practice time or require staff to create personal accounts without company approval.
An illustrative 180-minute agenda
- 0–15 minutes: establish the task, information boundaries and approval rule. Participants identify what must remain under human control.
- 15–35 minutes: the trainer works through one sample, including a mistake or missing fact. The group sees both the draft and the source.
- 35–65 minutes: participants attempt a similar task with guided support. Capture questions rather than replacing the exercise with more demonstrations.
- 65–80 minutes: compare outputs with the source. Discuss why a confident statement may still be unsupported.
- 80–90 minutes: break.
- 90–120 minutes: participants complete a fresh example independently. The trainer checks the reasoning and not just the final wording.
- 120–145 minutes: targeted feedback and a second attempt on the specific difficulty identified.
- 145–165 minutes: turn the method into a task template paired with a checking checklist.
- 165–180 minutes: identify the next practice task, responsible owner and questions that require separate follow-up.
This is an EverX editorial planning example, not a prescribed curriculum, a guaranteed learning duration or a published service commitment. Group size, starting level and task complexity may require more time or separate sessions. Ask a provider what gets removed if the schedule runs late; independent practice should not be the automatic casualty.
Use an exercise with an inspectable answer
Fictional notes: Mei will ask the supplier about availability. Ravi may review the draft, subject to his schedule. The team wants an update next week. No date or final owner for the update is agreed.
A flawed action list might assign Ravi as the confirmed owner and set Monday as the deadline. The expected response is not to accept a tidier list. It is to preserve Mei’s confirmed enquiry task, mark Ravi’s role as provisional and request clarification on the update owner and date. No real employee records are needed to teach that distinction.
Assess four observable behaviours
- Input selection: the participant uses the intended source and does not add unapproved information.
- Source fidelity: the output preserves what the source actually says.
- Uncertainty: missing facts stay visible instead of becoming invented commitments.
- Approval judgement: the participant can explain what is ready, what needs revision and who decides.
For each behaviour, mark independent, needs prompting or not yet demonstrated. Use the result to plan support, not to imply an accredited qualification. A participant can produce a good-looking answer by chance; asking them to explain the decision gives you more useful evidence.
IES recommends interleaving worked examples with problems and asking explanatory questions. We adapt those general teaching principles here; they do not validate this agenda or establish a specific corporate return.
Source: IES, Organizing Instruction and Study to Improve Student Learning, recommendations 2 and 7
Plan the week after the workshop
Give the team one current version of the template, one checking guide and a named person for unresolved questions. NIST includes personnel training within AI risk management, but training is only one part of an organisation’s responsibilities.
Source: NIST AI RMF, GOVERN 2.2
Agree one follow-up practice task and review the actual attempts. Satisfaction feedback can tell you whether the session was clear and well run; it cannot establish that the method survived contact with daily work. A realistic proposal distinguishes that follow-through from the initial session and states whether it is included.