AI for personalised lesson plans: the instructor still calls the shots
Personalising lesson plans by class was always a good idea — and a time nightmare. In 2026, AI turns this into a 15-minute job per class. Without replacing the instructor.
An experienced instructor knows that the ideal lesson plan depends on the class in front of them. A career-shift group wants practical cases; a beginner group wants fundamentals; a corporate group wants direct application to their sector.
The problem was always time. Personalising each plan to the class takes hours — and the instructor didn’t have those hours. Result: the template plan is used everywhere, and the learning experience suffers.
What AI changes
It doesn’t change the content — that’s still designed by the instructor, the pedagogical coordinator, the curriculum author. What changes is the fit: with good input about the class, an agent can produce an adapted version of the plan in minutes.
Three simple inputs do most of the work:
- Class profile (average age, level, main motivation).
- Stated goals (why did they enrol? what do they want to show at the end?).
- Sector context, if applicable (client company, type of business).
Where AI gets it right and where it fails
Right:
- Reframing exercises with examples from the student’s sector.
- Adjusting density — more theory for some, more practice for others.
- Suggesting relevant references and case studies.
- Preparing adapted quizzes and assessments.
Wrong:
- Detecting human dynamics in the class (distracted group, stuck student, tension between two). That’s an instructor’s eye — and should stay that way.
- Choosing the pedagogical moment to introduce a hard concept. That’s a decision made by someone reading the room.
- Replacing the experience of teaching — AI prepares the material, but class is still given by someone who knows.
The rule we use
AI drafts, instructor approves. Nothing goes to class without passing through the eyes of whoever teaches. In the first weeks, "approve as-is" is ~60%; after two months, with the model calibrated to the school’s style, it climbs to ~85%.
Time saved is real: a good instructor who used to spend 3-4 hours per class personalising plans now spends 30-45 minutes — reviewing and deciding what goes to class.
What a school has to sort first
Before hooking AI to lesson planning, three things must be tidy:
- Written curriculum. If there’s no base document for the course, AI will invent — and that’s where disasters begin.
- Documented school style. A creative school doesn’t write like a technical one. Writing this once saves a thousand corrections later.
- Recorded feedback. The instructor should tag "this yes" or "this no" on every review. Without feedback, the system doesn’t improve.
Once done, AI becomes assistant — and personalisation stops being a luxury to become the default.