AI Prompts for Instructional Designers – 25 That Actually Work

Most people's first AI prompts look like search queries: "Write a course on time management" or "Make a quiz about safety." The results are generic, somewhat bland and more work to fix than they were worth. Then a colleague gets a nearly finished storyboard from the same tool, and the difference is almost always the prompt. Good AI prompts for instructional designers carry the context a model needs to do real design work. Which prompts work, and what makes them better than the rest?

The AI prompts that work for instructional designers give the model a role, the audience, the goal, the source material, the format you want and the constraints to follow. That's the same briefing you'd give a capable new teammate. Below are 25 prompts organized by project phase, from analysis through evaluation, each written so you can copy it, swap in your details and get a usable draft. They work in ChatGPT, Claude, Microsoft Copilot, Gemini and most other assistants. Every one of them produces a first draft rather than a final product, so plan a review step for each.

What Makes AI Prompts for Instructional Designers Work?

Prompts work when they remove guesswork, because the model fills every missing detail with an average assumption, and averages produce bland training. The table below shows the components of a strong prompt.

In the prompts below, anything in square brackets is a placeholder for your own details. Always paste real source material when you have it, since grounding the model in your documents is the single most important step toward accurate output.

Analysis Prompts (1 to 5)

Analysis prompts help you understand the problem and the content before you design anything. They're especially useful for digesting long documents and messy SME notes.

1. "Read the procedure below. List every task a [job role] performs, in order, and mark which steps require judgment and which are routine. Flag any step that seems unclear or incomplete."
2. "Here are notes from an SME interview. Summarize the key tasks, the most common mistakes new employees make and the consequences of each mistake. Put it in a table."
3. "I'm meeting a stakeholder who asked for training on [topic]. Write ten questions to find out whether training is the right solution or whether the problem is really a process, tool or motivation issue."
4. "Compare these two documents and list every place where they disagree or where one includes a step the other doesn't."
5. "Based on this job description and these performance notes, create a learner profile: background, typical day, tech access, likely frustrations and what would motivate them to take this training seriously."

Prompt 3 is a favorite because it encourages performance consulting before anyone commits to building a course that might not solve the real problem.

Design Prompts (6 to 10)

Design prompts turn analysis into structure: objectives, outlines and practice ideas. Pair them with a solid guide to writing measurable learning objectives so you can judge the drafts quickly.

6. "Write four measurable learning objectives for [audience] based on the task list below. Use observable action verbs. Avoid 'understand,' 'know' and 'learn.' Note which level of Bloom's taxonomy each one targets."
7. "Propose three different ways to structure a [length]-minute course on [topic] for [audience]. For each, give the module sequence and one strength and one weakness."
8. "For each objective below, suggest one realistic practice activity that lets the learner do the task, not just read about it. Describe what the learner does and what feedback they get."
9. "Create a branching scenario outline where a [role] faces [situation]. Include three decision points, three choices at each point and the realistic consequence of each choice."
10. "Review this course outline and identify content that's nice to know rather than need to know. Suggest what to cut or move to a job aid."

Content and Scripting Prompts (11 to 15)

These are the classic ChatGPT prompts for instructional designers, and they work equally well in any capable assistant. They save the most hours, because scripting is where the typing accumulates.

11. "Using the outline and source below, draft storyboard screens 1 to 6 in a table with columns: screen title, narration (under 75 words), on-screen text (short bullets), visual description, interaction and developer notes."
12. "Rewrite this narration to sound conversational and natural when read aloud. Use short sentences and contractions. Keep the meaning exactly the same."
13. "Turn this policy section into a realistic workplace example that a [role] would recognize. Keep it under 120 words and end with a question for the learner."
14. "Write an engaging two-sentence opening for a module on [topic] that connects to a real frustration [audience] faces on the job. Give me five options."
15. "Create a one-page job aid from this procedure: numbered steps, short sentences, warnings clearly marked and no extra explanation."

Tip: after any content prompt, follow up with "What did you assume that wasn't in my source material?" It's a quick way to catch invented details before they reach your SME.

Assessment Prompts (16 to 20)

Assessment prompts produce questions, distractors and feedback. They work best when you specify the level of thinking and require a source citation for each answer.

16. "Write five scenario-based multiple-choice questions for this objective: [objective]. Four options each, one correct answer, distractors based on common real mistakes, similar option lengths. No 'all of the above.' Cite the source line for each correct answer."
17. "For each option in these questions, write one sentence of feedback explaining why it's right or wrong."
18. "Review these quiz questions for problems: more than one defensible answer, clues in the wording, unequal option lengths or questions that only test memorization. List the issues by question number."
19. "Create three variations of each question below so I can build a randomized question pool. Keep the same objective and difficulty."
20. "Write a short rubric for evaluating a learner's written response to this scenario: [scenario]. Include three criteria with descriptions of strong, acceptable and weak responses."

Accessibility, QA and Evaluation Prompts (21 to 25)

The last group handles the finishing work that's easy to rush: accessibility, quality checks and evaluation.

21. "This image appears on a screen teaching [concept]. Write alt text under 125 characters that conveys what the learner needs from it. If the image is decorative, say so."
22. "Rewrite this paragraph at about a grade 8 reading level without changing its meaning. Then list any terms that still need definitions."
23. "Compare the narration and on-screen text for each screen below. List any screen where they contradict each other, use different terms or where on-screen text repeats narration word for word."
24. "Check this storyboard for inconsistent terminology, spelling variations and style issues against these rules: [style rules]. Return a list by screen number."
25. "Here are anonymized open-ended survey comments from learners. Group them into themes, list the comment numbers that fit each theme so I can check the counts, and suggest one course improvement per theme."

What Should You Say When the First Draft Misses?

When a first draft misses, give specific feedback in the same conversation instead of starting over. The model retains the context, so a short correction usually gets you most of the way there, and while vague feedback like "make it better" rarely helps, specific feedback almost always does.

Here are follow-up lines worth keeping handy:

• "Too long. Cut each item by a third without losing the key action."
• "This sounds like a textbook. Rewrite it the way a friendly supervisor would explain it on the floor."
• "The examples are too generic. Use situations a [role] at a [type of workplace] would actually face."
• "You added details that aren't in my source. Remove anything not supported by the text I pasted."
• "Show me two alternative versions: one more formal, one more casual."
• "Explain your reasoning for this structure in three bullets."

The request for reasoning is the one people skip, and it pays back. Asking the model to explain its choices often reveals an assumption you disagree with, and it gives you language to use when you explain the design to a stakeholder.

Which Prompting Mistakes Waste the Most Time?

The mistakes that waste the most time are asking for too much at once, skipping the source material and trusting polished output without checking it.

Asking for a whole course in one prompt produces something long, shallow and difficult to review. Break the work into stages (analysis, then objectives, then outline, then screens), because each stage gives you a checkpoint where you can correct course before errors spread.

Skipping source material forces the model to rely on general knowledge, which may be outdated or simply different from how your organization does things. Fluent writing can also conceal incorrect details, so read every draft as if a new teammate wrote it, appreciating the effort while verifying the facts.

How Do You Turn These Prompts into a Team Habit?

Turn prompts into a habit by saving them in a shared library, improving them after each project and attaching each one to a review step. A prompt library is one of the most valuable assets an L&D team can build, because it captures what works and makes it repeatable.

• Store prompts in a shared document or wiki, grouped by phase.
• Add a "last updated" note and an example of good output for each.
• Pair each prompt with its review checklist.
• Invite the team to suggest improvements after every project.

Prompts also depend on the tool. Some work better in an assistant that handles long documents, others in the AI features inside your authoring tool. If you're still choosing a toolkit, our guide to AI tools for eLearning compares the main categories and where each fits.

One more habit worth building: when a prompt gives you a poor result, don't just correct the output; correct the prompt. Ask yourself what information the model lacked, add that to the prompt and save the new version, so that over time your library becomes sharper and your first drafts get closer to final.

Better Prompts, Better Drafts: Making These AI Prompts Your Own

The best AI prompts for instructional designers aren't magic phrases. They're good briefings: role, audience, goal, source, format and constraints. The 25 prompts above give you a starting point for every phase of a project.

Choose three that match your current work and try them this week, then adjust them to your audience and style, save the versions that work and share them with your team. A month from now, you'll have a prompt library that turns the blank page into a solid first draft far more often than not, and more time for the design decisions only you can make.

← All articles