Plan an AI Workshop Your Team Can Use the Next Day

Choose one useful learning outcome

“Learn about AI” is too broad to guide a session. A more useful outcome is: “Participants can draft a reply from an approved FAQ, check every factual claim, and identify when a person should respond.” The outcome gives the facilitator a way to choose examples and the team a way to know whether the session helped.

Ask participants what they currently do, where they get stuck, and how much experience they have with AI tools. A mixed group may need a short orientation before attempting a work-specific exercise. Do not assume that confidence with a chatbot means someone understands privacy, verification, or task boundaries.

Prepare safe practice materials

Create an invented business scenario or use public materials you have permission to reuse. Remove customer details, staff records, account information, and private documents from exercises unless their use has been specifically approved in the right environment. A live workshop is the wrong place to discover that an example contains confidential information.

Make sure the participants can access the approved tool before the session. Provide a non-AI version of the exercise if an account, accessibility need, or connection prevents participation. The learning goal should survive a temporary product outage.

Use a short explain, practise, review cycle

Begin with a brief demonstration using the same material participants will receive. Explain the task, the information supplied, the desired output, and the checks that follow. Then let participants do the work, compare different results, and discuss why those results vary.

For a 90-minute session, an illustrative agenda is 15 minutes of orientation, 15 minutes of demonstration, 30 minutes of paired practice, 20 minutes of review, and 10 minutes to choose the next step. This is a planning example, not a promise about any particular AI Empower course. Adapt the pace to the group and provide breaks where needed.

Make verification part of the exercise

Give participants an answer containing one unsupported claim, one missing detail, and one correct point. Ask them to identify each and locate the evidence in the supplied material. This makes checking a visible skill rather than an instruction everyone nods at and forgets.

A useful prompt pattern is: “Use only the material below. Draft a concise answer for this audience. List any missing information separately. Do not invent names, prices, dates, or outcomes.” This does not guarantee accuracy. Participants still need to compare the output with the source and decide whether the task is appropriate for AI.

The NIST AI Risk Management Framework offers a reference for discussing risks and responsibilities. Workshop exercises should make those responsibilities concrete in the team’s own workflow.

Leave with a small operating agreement

• Which tools and data are approved for the task?
• Who checks the output before it is used or sent?
• What kinds of work are out of scope?
• Where should a mistake, uncertainty, or privacy concern be reported?
• Who will maintain the shared example and checklist?

Keep the agreement short enough to use. Give the team one worked example, one verification checklist, and a place to ask questions. These are more useful than a long prompt library with no explanation of when a prompt is safe or effective.

Check what changed after the session

Ask participants to repeat the same bounded task later and describe what they checked. Compare quality, time spent, and corrections with the earlier attempt. Do not claim productivity gains from enthusiasm scores alone. If the practice is not being used, find out whether the task, tool access, or approval process is the obstacle.

Explore AI Empower’s workshop catalogue and bring the audience, current skill level, desired outcome, and examples of the work. A focused session can connect learning to a practical pilot without promising promotions, funding, or guaranteed business results.