Lesson Preparation with AI: 6-Step Workflow plus Prompt Templates

Teachers in the Netherlands spend an average of 8-12 hours per week on lesson preparation (Onderwijsraad, 2022). AI can take over a large part of that — not by taking over the work, but by accelerating it where it is dull and repetitive. In this article: a 6-step workflow, concrete prompt templates for 5 lesson types, and — just as important — what AI does and does not do for you and what that means for your role as a teacher.

What AI does and does not take over

The biggest misconception in AI lesson preparation: assuming AI writes the entire lesson for you. That's not true, and it's not what you want either. Below is the dividing line:

What AI does

• Initial outline of slide structure based on learning objectives
• Generating quiz questions at different Bloom levels
• Coming up with case studies that apply theory
• Generating examples and analogies for abstract concepts
• Compiling summaries for students
• Translating learning objectives into concrete learning activities

What AI does not do

• Deciding which learning objectives are relevant (curriculum context, BPV connection, level mix)
• Content check for subject correctness (you remain ultimately responsible)
• Assessing class dynamics and which working method suits your group
• Anecdotes from your own practice that bring the material to life
• Adjusting during the lesson based on what you see in the moment
• Relationship with students — trust, humor, individual attention

Translation: AI does the structure and bulk content; you do curation, didactic choices, and human connection. That's the division of labor that yields a 50% time gain.

The 6-step workflow

Step 1 — Formulate 1-3 concrete learning objectives

This is your job, not AI's. A learning objective must be specific and measurable. Bad: "Students understand market forces.". Good: "The student can apply supply and demand dynamics to a current practical example from the retail sector.".

Three learning objectives per 50-minute lesson is a good rule of thumb. More than four is too much cognitive load for one lesson block. Time investment: 5-10 minutes.

Step 2 — Write a prompt with learning objectives + context

The quality of AI output is directly related to the quality of your prompt. Good prompt elements:

• Subject and level: "MBO-V level 3, subject health education"
• Target group specification: "Class of 22 students, mixed in terms of language proficiency"
• Learning objectives literally: Paste your learning objectives from Step 1
• Lesson type: Theory (introducing knowledge), Case (applying) or Quiz (testing)
• Length: "50-minute lesson with 3-4 interaction moments"

Example prompt: "Generate a 50-minute health education lesson for MBO-V level 3. Class of 22 students, mixed language proficiency. Learning objective: the student can recognize symptoms of dehydration in the elderly. Lesson type: Theory + Case Flow. Add 3 interaction moments and end with a summary."

In LectaMe, you do this via the generator wizard (no typing prompts, but input fields). Time investment: 2-3 minutes.

Step 3 — Generate and read critically

Click generate, get a first output (30 seconds in LectaMe). Read everything for:

• Subject correctness: Is every claim correct? AI sometimes hallucinates facts. Check especially statistics, years, and specific definitions.
• Learning objective coverage: Is every learning objective actually addressed, or has it slipped into side issues?
• Level passing: Is the language not too formal for N3? Not too superficial for HBO?
• Quiz question quality: Apply the 8 rules from our quiz article (see quiz making).

Time investment: 10-15 minutes. This is the crucial phase — here you prevent AI errors from entering your classroom.

Step 4 — Adjust, delete, add

Edit the output. Three typical edits:

• Add your own anecdotes. An example from your own practice makes a lesson memorable. AI does not know your stories — add them manually to the right slide.
• Replace superficial examples. AI sometimes gives generic examples ("a patient comes in"). Replace with specific images from the BPV context of your students.
• Delete excess. AI is generous with content. For 50 minutes, you need fewer slides than AI initially makes. Deleting improves focus.

Time investment: 10-15 minutes.

Step 5 — Place interaction elements

Add a quiz question or poll per concept. Follow the pattern from our article active working methods: level-1 question directly after the introduction of a concept, level-3 application question halfway, level-4 or 5 synthesis question at the end.

LectaMe's AI often already suggests interaction moments — accept them selectively, add where necessary. Time investment: 5-10 minutes.

Step 6 — Test on one colleague or test target group

Optional but powerful: have a colleague quickly review the lesson. Ask: "do you understand where this is going?" Not "is this good?" — that gives socially desirable answers. "Do you understand it?" gives honest feedback.

Time investment: 5 minutes review time from colleague.

Total time investment versus traditional

Step	With AI (Lectame)	Without AI (classic)
Learning objectives	5-10 min	10-15 min
Slide setup	2-3 min (prompt)	60-90 min
Creating quiz questions	Included in AI output	20-30 min
Creating case studies	Included in AI output	20-30 min
Critical review	10-15 min	5 min (own work)
Editing / supplementing	10-15 min	30-45 min
Placing interactions	5-10 min	15-20 min (manual setup)
Colleague check (optional)	5 min	5 min
Total	40-60 min	165-235 min

Conclusion from this table: AI-supported preparation saves 60-70% time for a comparable lesson setup — provided you take the review phase seriously. If you skip it, the AI output will go unfiltered into your lesson and that's where the problem arises.

Prompt templates for 5 lesson types

Template 1 — Theory lesson (introducing knowledge)

Generate a [DURATION]-minute lesson for [SUBJECT] at level [LEVEL].
Target audience: [TARGET AUDIENCE DESCRIPTION].
Learning objective: [LEARNING OBJECTIVE].
Lesson type: Theory Flow with progressive setup.
Including: 1 concept check after each concept, 1 summary at the end.
Show: [FORMAL/INFORMAL].

Template 2 — Case lesson (applying)

Generate a [DURATION]-minute case lesson for [SUBJECT] at level [LEVEL].
Target audience: [TARGET AUDIENCE].
Learning objective: [LEARNING OBJECTIVE applied].
Lesson type: Case Flow with 2-3 practical case studies.
Case studies should vary in complexity to allow for differentiation.
BPV context: [TYPE STAGE if relevant].

Template 3 — Quiz lesson (testing + repetition)

Generate a [DURATION]-minute repetition lesson for [SUBJECT].
Topics to test: [LIST 3-5 TOPICS].
Lesson type: Quiz Flow with a mix of level-1 and level-3 questions.
Per topic: 1 concept question + 1 application question.
End with an open reflection question.

Template 4 — Workshop / training (business)

Generate a [DURATION]-minute training on [TOPIC].
Target audience: [PROFESSIONALS in which role].
Learning objective: [BEHAVIORAL OBJECTIVE — what should they do differently tomorrow?].
Style: Case Flow with scenarios from practice.
Including: interactive poll on current work method, scenario case,
final action questions ("what one thing will you do tomorrow?").

Template 5 — Differentiation lesson with layers

Generate a [DURATION]-minute lesson for [SUBJECT].
Target audience has level differences: [E.G. N3 AND N4 TOGETHER].
Lesson type: layered teaching — basic for everyone + in-depth layers.
Per concept: basic layer + 1 in-depth layer.
Quiz questions at different levels (see rule: basic for all,
in-depth for those who have seen that layer).

4 quality checks for AI output

For every lesson you create with AI, perform these 4 checks before presenting:

Check 1 — Hallucination check

Verify all facts, years, and statistics. AI sometimes invents specific numbers ("78% of patients experience X"). If in doubt: delete or verify in a source.

Check 2 — Learning objective coverage

Go back: does this lesson cover each learning objective from Step 1? Remove content that does not support a learning objective — no matter how interesting it is.

Check 3 — Cognitive load check

Too many concepts in one lesson overwhelm. For 50 minutes: max 3-4 new concepts. Count them in your AI output — is it more? Remove or divide over multiple lessons.

Check 4 — Own voice check

Read the text aloud. Does it sound like you? Or does it sound like generic AI sentences? Replace formulations that do not fit you — otherwise, students will feel the difference between your voice and that of the AI.

Common mistakes in AI lesson preparation

• Too vague prompts. "Create a lesson about diabetes" yields something generic. Provide level, target group, learning objective, lesson type, duration.
• Blindly accepting output. Without critical review, AI errors enter your classroom. Build in 10-15 minutes of review time.
• Not adding your own anecdotes. An AI lesson without a teacher's voice feels clinical. Add at least 1 personal example per lesson.
• Leaving everything to AI, including learning objectives. AI does not know your curriculum. Learning objectives remain a teacher's task.
• Not reserving time for repeat iterations. The first AI output is rarely your final product. Plan time for 1-2 revised rounds per lesson.

What this means for your role as a teacher

AI changes your role as a teacher — but not as techno-optimists and techno-pessimists claim. No replacement, no dehumanization. Rather, a shift:

• Less time spent on structure building, more on didactic choices. AI does the slide structure; you determine which working method, which level, which emphasis.
• More time spent on class dynamics during the lesson. With 60% less preparation time, you can pay more attention to what happens in the moment.
• More quality control as a responsibility. You become the final editor of AI output. That is a new skill — critically reading for level, subject correctness, and didactic fit.
• Less repetitive work, more creative work. AI is good at the 'first 80%'. The last 20% — the creativity, the finesse, the adaptation to your class — is exactly where you add value.

Conclusion

AI lesson preparation is not laziness and not dehumanization. It is division of labor: AI does structure and bulk, you do curation and human connection. With the 6-step workflow, you gain 60-70% preparation time, provided you take the review phase seriously.

Don't start with your most important lesson. Take a second-priority lesson and test the workflow there. The first time it takes about as much time as doing it manually — from the third or fourth lesson you'll experience the time difference.

Want to try it out right away? Generate a free lesson with LectaMe — fill in subject, level, and learning objective and go through Steps 3-5 on the output.

Read on

• Creating quizzes — 8 rules for strong quiz questions
• 10 active learning methods with research
• What is layered teaching? (Template 5)
• For teachers — all LectaMe features
• Features — what the generator can do

Sources

• Education Council (2022). Teacher shortage and workload in Dutch education. The Hague.
• Hattie, J. (2009). Visible Learning. Routledge.
• Mayer, R. E. (2009). Multimedia Learning (2nd edition). Cambridge University Press — cognitive load principles.
• Biggs, J. & Tang, C. (2011). Teaching for Quality Learning at University — constructive alignment behind the Theory/Case/Quiz Flows.
• Selwyn, N. (2019). Should Robots Replace Teachers? Polity — on what AI in education can and cannot do.