Free online workshop for instructional designers · AI for IDs
Three reasons your AI output is bad
You wrote a careful prompt and got back something you could not use. Usually one of three things went wrong: the model was the wrong tier for the task, the instruction did not carry the criteria the model needed, or nobody checked the output against a standard before it went out. In 60 minutes we cover all three with real examples across ChatGPT and Claude, then you fix one of your own prompts live.
- When Saturday, September 26, 2026 · 9:30 AM Eastern
- Length 60 minutes
- Where Online, link sent by email
- Cost Free
What you leave with
Three things you can use on Monday
-
1 · Models
Which model your task actually needs
The two-chat test that tells you in minutes whether you are on the wrong tier, and the tells that give it away: one-line questions that come back as essays, answers confident about half of what you asked.
-
2 · Instructions
Why the same prompt fails on a different model
Mid-tier models want tight criteria. Top-tier models want the outcome and a question back. What that difference looks like in a real prompt, side by side, on ChatGPT and Claude.
-
3 · Evaluation, live
Fix one of your own prompts on the call
Run the goal, criteria, evidence prompt on a task you brought. Check the output against your own criteria. Name one specific fix. You leave with a template you built, not one we handed you.
- Format30 minutes of content, 20 minutes of Q&A, 10 minutes of live fixes. Recorded, and every registrant gets the recording.
- Who it is forInstructional designers, curriculum developers and L&D teams who already use AI at work and are not happy with what comes back.
- Who runs itThe 24/7 Teach instructional design team, the people who teach our AI-age instructional design bootcamp.
Bring one thing
One real task, and one prompt that gave you output you could not use. That is what you fix in the live segment. Use non-confidential material.
Registration
Save your seat
Name and email is all we need. You will get a confirmation now, the join link before the session, and the recording afterward. Nothing else unless you ask.
You're registered.
Saturday, September 26, 2026 at 9:30 AM Eastern. A confirmation is on its way to your inbox.
- The join link comes by email before the session.
- Bring one real task and one prompt that gave you output you could not use.
- Know a colleague who fights with AI output? Send them this page.