Building an online course takes 200-400 hours if you do it manually. Researching modules, building lesson outlines, writing scripts, creating workbooks, designing assessments. It's a project.
With AI, you can build a complete, functional course outline in 3-4 days. Not a rough outline. A detailed, module-by-module, lesson-by-lesson breakdown with scripts and assessments ready for recording.
This guide walks through the exact workflow: from "here's my course idea" to "here's a production-ready curriculum." The tools that work (Claude for quality, ChatGPT for speed). The prompts. How to structure module progression. How to write effective lessons. The role of assessments and how AI helps build them.
Part of our complete AI writing guide. See also tool reviews and prompt engineering.
Real talk: AI is exceptional at course scaffolding and structure. It's mediocre at deep expertise and unique methodology. Build the structure with AI. Fill the substance with your knowledge.
The Course Building Workflow: From Idea to Curriculum
Phase 1: Course Conceptualization (30 minutes)
Define your course in one sentence. Who's the student? What problem do you solve? What's the outcome? What's the prerequisite knowledge?
Write these down. They're your inputs for AI.
Phase 2: Module Structure Generation (30 minutes)
Feed your course concept to Claude (Claude is better for this than ChatGPT). Ask for a module breakdown.
The Prompt:
"Design a comprehensive course about [topic]. Target student: [describe]. Learning outcomes: [what they'll be able to do]. Duration: [how long]. Create 4-6 modules that progress logically. For each module, list: 1) Module title and goal, 2) 3-5 lessons within the module, 3) estimated duration per lesson, 4) type of content (video/interactive/written), 5) assessment method. Format as a structured outline."
Phase 3: Lesson Script Generation (2 hours)
For each lesson, generate a script. This is where you switch between ChatGPT and Claude. ChatGPT is faster for quantity. Claude is better for quality if you're only writing 2-3 scripts.
The Prompt (per lesson):
"Write a lesson script for [lesson title]. Module goal: [module goal]. Prerequisite knowledge: [what they should know]. Learning objective: [what they'll learn]. Duration: [8 minutes / 15 minutes]. Format: video script. Include: 1) Hook (30 seconds), 2) Problem setup (2 minutes), 3) Core content (6 minutes), 4) Example (2 minutes), 5) Takeaway (1 minute). Tone: [friendly/authoritative/conversational]. Target audience: [describe]."
Phase 4: Workbook and Assessment Creation (1-2 hours)
Generate worksheets, exercises, and quizzes for each module.
The Prompt:
"Create a workbook module for the [module title] course module. Students just learned about [lesson topics]. Create: 1) 3-5 fill-in-the-blank exercises reviewing key concepts, 2) 2 applied exercises where they use what they learned, 3) a short self-assessment quiz (5 questions), 4) one 'challenge problem' that stretches them slightly beyond what they learned. Include answer keys. Format as a PDF-ready document."
Phase 5: Edit, Polish, and Test (4-6 hours)
This is the human work. Read through everything. Does it flow? Does the progression make sense? Are there gaps? Does the workbook match the video lessons?
This phase takes time, but it's the phase that transforms AI scaffolding into a real course.
Module Structure: The Framework That Works
Most courses fail because the module progression doesn't make sense. Here's a structure that works:
Module 1: Foundations
Answer the question: "Why does this matter?" Share the problem you're solving and the stakes. Get them invested.
Module 2-4: Core Skills
Build skill sequentially. Module 2 introduces the foundational concept. Module 3 builds on it with application. Module 4 teaches the advanced version or handles edge cases.
Module 5: Integration
Show how everything connects. Real-world use. Case studies. How to combine what they learned.
Module 6: Mastery
Advanced projects, challenges, or real client/student work they can do with their new skills.
Ask Claude to generate your modules using this framework. It works better than asking for "a course outline" because it has structure baked in.
The Lesson Template
Each lesson should follow a pattern. AI is excellent at following patterns.
- Hook (30 seconds): Why should they care about this lesson?
- Problem (1-2 minutes): What's the challenge they're solving?
- Content (6-8 minutes): The meat. Teach the skill or concept.
- Example (2 minutes): Show it in action. Concrete example.
- Takeaway (1 minute): What's the one thing they should remember?
This template works for video, audio, or written lessons. Feed it to AI with your specific content, and it generates lessons that fit the pattern automatically.
Assessment Strategy
Assessments serve two purposes: 1) Check if students learned, 2) Give them confidence they can apply what they learned.
Ask AI for three assessment types per module:
- Knowledge checks: Multiple choice, true/false. Do they retain the facts?
- Applied exercises: Here's a scenario. What would you do? Tests if they can apply the concept.
- Project or case study: Real-world scenario. Forces integration of everything they learned.
AI is great at generating knowledge checks and basic applied exercises. Project briefs need your judgment.
The Tools
Claude is better for course design because it produces more thoughtful, coherent structures. ChatGPT is faster if you want to generate all lessons quickly and refine later.
Most creators use Claude for the outline (pays off in quality early) and ChatGPT for the individual lesson scripts (pays off in speed and iteration).
Course Building Checklist
- Day 1: Conceptualization + Module structure (1 hour)
- Day 2: Lesson script generation for modules 1-2 (4 hours)
- Day 3: Lesson scripts for modules 3-4 (4 hours)
- Day 4: Workbooks, assessments, editing (6 hours)
- Day 5+: Record, test, refine (ongoing)
With this workflow, you have a production-ready curriculum outline in a week. Recording and refinement takes longer, but the scaffolding is done.
The Real Advantage
The biggest win: you can finally build that course you've been thinking about for two years. The scaffolding work that kept you stuck is gone. You're left with: does the content work? Is this actually valuable? Can students apply it?
For more on AI writing workflows, see YouTube scripts, blog posts, and prompt engineering.
