Which tool to choose
To build the content a generalist assistant will do: ChatGPT and Claude both write programs, examples and quizzes. Claude handles long, coherent materials better; the versions with web search help you verify a figure or a regulation you cite. For the slides, have these AIs give you the outline and the text of each slide, then lay it out with the presentation tool you already use.
How to do it
On computer or on phone only where you then lay it out changes; the preparation is the same.
- Define the three boundaries. Topic, real duration (one or two sessions, how long) and the team's starting level. Training "on everything" teaches nothing; targeted training does.
- Ask for the complete package. The working syntax:
Act as a corporate training designer. I have to train my team on how to handle customer returns in two one-hour sessions. Create: 1) a program with clear learning objectives for each session; 2) for each point a simple explanation with a practical example tied to our work; 3) an exercise or a real case to have them work through; 4) a short check quiz with the answers. The team's level is basic. Our context: a small electronics shop that also sells online.
- Verify the facts. Procedures, numbers and rules cited must be checked: AI can confidently write a wrong detail. What you teach becomes the rule for the team, so it has to be correct.
- Add practice. Ask the AI to "increase the exercise part and reduce the theory": you remember what you do, not what you listen to.
A concrete example
Anna coordinates four people in a call center and has to train them on using a new management system. She isn't a trainer and fears boring them. She describes the system to Claude, the duration (one hour) and the team's basic level. The AI prepares her a program in three blocks, each with an explanation, an example taken from their real work (logging a complaint, updating a customer record) and an exercise to do together at the computer.
Anna verifies that the system's steps are exact, corrects two menu names, and adds a final quiz of five questions. The session becomes hands-on instead of a lecture: the team really tries while learning. By the end they know how to use the tool, not just what it does. Without AI, Anna would have improvised a spoken explanation and forgotten half the things.
When it does NOT work (and how to fix it)
If the content is generic and not about your case
Abstract training slides right off. Insist in the prompt on examples and cases taken from the team's real work, with their tools and their situations. The brain latches onto what it recognizes; it ignores what holds for anyone.
If there are errors in the materials
AI can invent a figure or a step in a procedure. Treat every technical fact as something to verify before the room: a wrong number taught to four people becomes four mistakes. For internal procedures, check every step yourself on the real system.
If it's all theory and little practice
If the session is a monologue, no one remembers anything the next day. Ask the AI to turn each concept into an exercise or a question to the group. Good training is made more of things the team does than of things you say.
A tip from someone who really uses it
Ask the AI also for a "one-page summary sheet" to leave with the team after the session. Training is forgotten quickly; a clear sheet to come back to, with the steps and the examples, is what makes what you taught stick. The course is the event, the sheet is the memory.
Frequently asked questions
How long should a training session last?
Better short, frequent sessions than a single marathon. Attention drops quickly: two one-hour meetings with an exercise each are worth more than half a day of slides. Ask the AI to break the topic into digestible blocks.
Can AI adapt the training to different levels in the team?
Yes. Tell it in the prompt ("some are beginners, others experienced") and ask it for two paths or exercises at different difficulty. That way those who already know don't get bored and those starting from zero don't fall behind.
Can AI run the course for me?
Here's the misunderstanding to clear up: AI prepares the material, but training happens in the exchange. The unexpected questions, the examples that arise from the group, reading from people's faces who hasn't grasped it — that's the heart of a course, and you lead it. The tool takes the preparation work off your hands; the room stays a thing between people.