Everyone can generate a course now. ChatGPT, Claude or Copilot will write you twelve modules before your coffee gets cold. But fast content is not the same as learning that sticks. Most AI-generated training looks polished, reads well and is forgotten within weeks. Not because the AI is bad, but because nobody told it what actually makes people better at their job.

In this 30-minute session, Margot Sprenkels (Training Optimisation Lead at aNewSpring) shows how to tame your LLM. You'll see what to feed it so it works like a learning designer instead of a copywriter: your sources, your L&D knowledge, clear instructions and examples of good and bad work. And you'll see the difference it makes, with the same lesson built twice: once without that input, once with it. Flat text turns into scenarios, practice moments and questions that test whether people can apply what they learned.

Better content is only half the answer, though. AI made content cheap, so what becomes scarce is evidence that people can do the job and keep doing it. That takes practice: rehearsing the real situation, more than once, with feedback. An AI role-play simulator, where learners take on a difficult conversation with a persona that responds like the real thing, is a strong way to get there. Your learning platform then brings that practice back at the right moment and shows who can actually do it. Content, practice, spacing and proof together turn training into behavior change. That's why AI won't replace your LMS. If anything, it makes it more important.

You'll leave knowing what to feed your AI on Monday morning, and what it can and can't do for you. And you won't leave empty-handed. Everyone who attends gets our free AI Workspace for L&D Starter Pack, the folder we work from ourselves: a paper with 21 evidence-graded principles of how people learn (a checklist for every course you build with AI), our trend report The Learning Economy, a project instruction template with a worked example, and five ready-to-use AI skills (from a meeting agent to a voice profile interview).

During this session, you will learn:

• The five things AI does when you give it nothing to work with: it flatters, it averages, it invents, it skips and it sounds right
• The four things to feed your AI before you ask for content: your sources, your L&D knowledge, project instructions and skills
• How to check what your AI gives back, and spot where it filled gaps your source never covered
• How to get AI to build practice, scenarios and questions that test whether people can do the job, not just remember it
• Where your AI stops and your learning platform takes over: practice over time, AI role-play to rehearse real conversations, spacing and evidence of skills