Summary: If you are shopping for AI classes for a team, the useful ones are built around the work your people already do, use the tools they will actually have, and end with something running. The weak ones tour features and leave. Here is how to tell the difference before you book, and when it is better to wait.
Why most AI classes do not stick
The usual pattern is a half-day session, a deck of examples from other industries, and a room of people who nod and go back to work the same way they did before. Nobody was careless. The class was just about AI, and your team’s job is not AI. It is quoting, scheduling, drafting letters, answering parents, writing job descriptions, or reconciling reports.
When the examples are not your work, people have to translate them on their own, and most will not. A class sticks when the first thing someone does with the tool is a task they would have done anyway that afternoon.
What to look for
- Your material, not a demo dataset. Ask whether participants will work on their own emails, documents, forms or spreadsheets. If the answer is “we provide sample files,” expect a nice day and little change.
- One or two tools, taught well. A survey of every chatbot on the market is entertaining and forgettable. Pick the tool that fits your existing software and data rules, then go deep.
- Role-specific practice. An office manager, a dispatcher and an owner do not need the same hour. Good training splits by role or at least lets each person bring their own tasks.
- A data rule people can remember. Before anyone pastes a client file into a chat window, they should know what is allowed. If the class does not cover this, you are creating risk, not skill.
- Something left behind. Saved prompts, a short checklist, a shared template. If everything lives in the instructor’s slides, it leaves when they do.
- Honesty about limits. AI gets things wrong, including names, numbers and citations. A class that does not show your team how to check the output is teaching overconfidence.
What to skip
- Certificates as the point. A badge does not tell you whether anyone does their job differently.
- “Prompt secrets” and magic formulas. Clear instructions, the right context and a habit of checking the answer cover most of it. If you want a quick read on your own prompts, the prompt checker is free.
- Hype about replacing staff. It makes people defensive and they stop experimenting.
- Training with no owner afterward. If no one is responsible for what happens in week three, week three will look like week zero.
Workshop or cohort: how to choose the format
The format is a scoping decision, not a different product. A short workshop suits a small team with a few clear tasks: you spend a few hours on real work and leave with working examples. A multi-week cohort suits a larger organization, where people need time to try something on the job, hit a snag, and bring it back. The second is slower and tends to produce more habit change, but only if the team has the time to practice between sessions.
Whichever you pick, the scope should come from the work: how many people, which roles, which tools, and how much prep is needed before the first session.
Before you book anything
- Write down the outcome. “Everyone uses AI” is not an outcome. “Estimates go out the same day” or “first drafts of parent emails take minutes” is.
- Settle your data rules. Even a one-page policy is enough to start. If you need one, the AI policy generator gives you a draft.
- Check readiness. The AI readiness quiz takes a few minutes and shows where the gaps are.
- Pick the tool. Decide which assistant your team will use so training and tools match.
Frequently asked questions
What should an AI class for a business team cover?
The tasks your team does every week, done with the tools they will actually use, on their own material. Tool tours and general theory matter much less than practice on real work, plus clear rules on what data can and cannot go into an AI tool.
Is a one-day workshop enough?
For a small team with a few clear use cases, often yes. For a larger organization with several roles, a multi-week format works better because people need time to try things on real work and bring back what broke.
Should we pick one AI tool before training?
Yes, if you can. Training people on three tools at once splits their attention. Choose the tool that fits your existing software and data rules, teach that one well, and widen later.
When is AI training not worth it yet?
When nobody has decided what the team is supposed to do differently afterward, or when you have no policy on what data can go into AI tools. Fix those first, or training becomes a one-off event that fades within a month.
When this matters - and when it doesn’t
This applies if: you lead a team of people whose daily work is writing, communicating, scheduling or analysis, and you want them using AI on that work with clear guardrails.
Skip it for now if: you cannot name one task you want done differently, or the team has no time to practice afterward. In that case a short AI Workflow Assessment to find the work worth changing will do more than a class would.
My own team training is built around your team’s actual work, not slideware. If you want to talk through whether a workshop or a cohort fits, you can book a call.
Want the step-by-step version?
The free guides walk through the prompts, workflows and checklists behind articles like this one, with no sign-up.
Browse the free guides