Key Takeaways
- Design the workshop backwards from the tasks people should be doing differently months later.
- Use real, approved documents and tools, and spend most of the session on hands-on practice.
- Plan follow-up before the day: short clinics, internal champions and a shared prompt library.
- Measure use and outcomes over months, not satisfaction scores on the day.
- Refresh materials as tools change instead of repeating the same workshop.
Why workshop enthusiasm fades
Plenty of organisations have run an AI session that went well on the day. People were engaged, the demos landed and the feedback forms were positive. A few weeks later, habits drift back to where they were. The content is rarely the problem. The workshop was treated as an event rather than the start of a change in how people work.
A workshop that sticks is designed backwards from the behaviour you want to see months later. That means choosing the tasks people will do differently, giving them enough practice to feel confident and putting support in place for the weeks when they get stuck. This article covers the three parts that matter most, design, follow-up and measurement, and draws on how Artifex runs team training.
Design around real work
Start by talking to a few people in each role that will attend. Ask what fills their week, which tasks feel repetitive and where they already use AI or avoid it. From those conversations, pick a handful of concrete tasks for each group. For a customer service team that might be summarising long complaint histories and drafting first replies. For analysts it might be cleaning data and explaining results in plain language.
Build the session around those tasks, using real, approved documents with sensitive details removed where necessary. Generic examples are easier to prepare but harder to carry back to the desk. When people practise on the kind of email thread or report they will handle on Monday, the connection to their own work is obvious. A few design rules help:
- Group attendees by role so every exercise is relevant to everyone in the room.
- Use the tools people are actually permitted to use, set up and tested in advance.
- Cover safe use early: what data can go where, how to check outputs and when not to use AI.
- Keep presentations short and give most of the time to exercises, with trainers circulating.
- End with each person writing down one task they will change in the coming week.
A simple session outline
- Context.A short explanation of what the tools do well and badly, illustrated with examples from the team's own work.
- Guardrails.The organisation's rules on data, accuracy and review, shown in practical terms rather than policy language.
- Guided practice.A trainer works through one real task step by step while everyone follows along on their own screen.
- Independent practice.Participants tackle their own tasks and compare approaches with colleagues, asking for help as needed.
- Commitments.Each person names a task to change, and the group agrees where and when they will share results.
Plan the follow-up before the workshop
The weeks after the workshop decide whether it sticks, so plan them before the session and tell attendees what is coming. The most useful follow-up is short and regular. A clinic where people bring a task they are stuck on is worth more than a second lecture.
Managers matter here more than trainers. If a team leader asks in the weekly meeting how people got on with the tasks they committed to, the new habits get attention and time. If nobody mentions them again, they quietly lose out to the usual workload. Brief managers before the workshop so they know what their teams will be trying and how to support it.
Identify a few internal champions in each team. They need not be technical, but they should be curious and trusted by colleagues. Give them extra coaching and a little time each week to help others. Set up a shared place for prompts, templates and examples that worked, so good ideas spread instead of staying on one laptop. Our AI enablement programme formalises this with champion coaching and regular office hours.
The workshop teaches the skill. The months that follow decide whether people think it is worth keeping.
Measure what changed, not how people felt
Feedback forms on the day tell you whether the session was enjoyable, not whether anything changed. Decide in advance what you will look at, and check it at intervals over the following months, for example a few weeks after the session, again at three months and again at six.
Useful measures include whether people still use the approved tools, whether the tasks they committed to are being done differently and whether those tasks now take less time or produce fewer errors. A short confidence self-assessment before the workshop and again later also shows progress. If you mapped the work beforehand, compare against that baseline, as described in our five-step audit.
If progress stalls, ask why. Sometimes the tool does not fit the task, sometimes people lack time to practise and sometimes a rule blocks a sensible use. Each has a different fix, and the answer is rarely another identical workshop. Talk to a few people who stopped using the tools as well as those who carried on, because the reasons they give are usually more useful than any dashboard.
Keeping it current
AI tools change quickly, and so do the tasks people apply them to. Review the materials and the prompt library every few months. Retire what no longer works, add examples from teams that have found new uses and run short refreshers for groups whose tools have changed. Treat the workshop as the first step in a programme, and people will still be using what they learned long after the day itself.