Choose one repeatable task
Ask the team to identify a recurring task with a clear output: turning fictional meeting notes into an action list, drafting a routine update or organising ideas for a presentation. Choose something participants understand well enough to review. A task with an unknown correct answer is a poor starting point for learning how to check AI output.
Write down the inputs, expected output and person who normally reviews the work. Record any steps that require a human decision. The guide to choosing AI training can help you compare courses around those tasks rather than around a long list of product names.
Check tools and access before the workshop
Ask your organisation which tools and accounts participants may use. Confirm the current workshop requirements with the provider, including devices, browsers, logins and any subscriptions. Product access and features change, so an old screenshot or a colleague’s account should not be treated as proof that everyone will have the same functions.
- Can each participant sign in using an approved account?
- Does the training network allow the required service?
- Who can help with access problems on the day?
- Is there a suitable alternative exercise if a tool is unavailable?
For application-building workshops, confirm installation and device requirements separately. A writing exercise and a development exercise may require different preparation. Compare AI Productivity Playbook with Vibe Coding 101 to understand the intended learning tasks.
Bring a sample the team is allowed to use
Start with fictional information or an example explicitly approved for the workshop. Remove details that are not needed for the task. A simple practice brief can be useful without including real customer records, employee information or confidential business plans.
Fictional notes: “Amina will draft a short team update by Thursday. Raj will check the figures; the deadline is still undecided. The group will review the draft next week, but no meeting date has been agreed.”
Expected output: An action register that preserves the named owners and agreed deadline, and clearly labels the missing deadlines and meeting date.
Agree how a human will check the result
Use a checklist before deciding whether the output is useful: does it preserve the source facts, distinguish unknown information, meet the requested format and suit the audience? Ask the reviewer to identify an error and explain a correction, rather than simply say that the output looks good.
Plan a second attempt. Change one instruction, compare the outputs and record what improved or became less reliable. Measure the whole task, including preparation and review, if you want to compare effort. A fast first draft can still require substantial checking.
Prepare a realistic follow-up
Choose one approved workplace task for participants to try after the workshop. Name the reviewer and agree how questions will be handled. Use a small pilot to understand the process before assuming that the same approach will suit every task or department.
Download the AI workshop readiness worksheet (.txt). Connect the follow-up to a simple evaluation plan and read Yavinesh Rajagopal’s trainer profile. Share the team’s chosen task through the training enquiry form so the provider can advise on a suitable programme.