Case study · Curriculum design
Building a coherent AI curriculum, from Bachelor's to Master's.
The starting point
AI content isn't in short supply. What's missing, over a long programme, is a progression: a Bachelor's student and a Master's student don't share the same background, the same professional expectations, or the same career horizon.
Without an overall structure, two flaws appear almost automatically. The same fundamentals get repeated year after year, and advanced skills never arrive because nobody has decided when to introduce them.
A progression by level, not a stack of courses
The programme was built from the target skills at graduation, then worked backwards to the entry point. Each level receives what it can realistically absorb, and explicitly prepares the next one.
That work forces sharp trade-offs: what to leave out at Bachelor's level so it can be done properly at Master's, and conversely what shouldn't be pushed back at all.
Content anchored in professional use, not theory
Each module pairs course content, hands-on projects and assessment, built on real professional uses of AI rather than a theoretical progression.
The goal fits in one sentence: a graduate should not only know how to use the tools, but be able to judge what they produce and explain their choices to a non-technical leadership team.
Building an agent, not just talking about one
The most structuring module of the curriculum has students build an automated agent end to end on n8n. Not a demo watched in class: each student assembles their own flow, triggers it, breaks it, and fixes it.
The choice of n8n is deliberate. The tool makes visible what an agent actually does, step by step: the trigger, the call to the model, the tool the agent uses to retrieve information, then the resulting action. A student who has wired those pieces together understands what sits behind the word "agent", and can then spot what's off in a sales pitch.
The exercise also works as a teaching device. It's while connecting a knowledge base to an agent that students discover, concretely, why data quality matters more than model choice.
The agent module, built on n8n
How Next-U was supported
Target skills at graduation
What a graduate needs to be able to do, framed as verifiable actions rather than knowledge.
Breakdown by level
Distributing skills from Bachelor's to Master's, avoiding both repetition and gaps.
Content and hands-on projects
For each module, the course, the exercises and the cases worked on real situations.
Assessment design
Assessments that measure use and critical judgement, not the recall of definitions.
What it delivered
- A coherent AI programme across the whole curriculum, from Bachelor's to Master's.
- An explicit progression: each level knows what it must deliver and what it hands to the next.
- Content, hands-on projects and assessments built on real professional uses.
- A module where every student builds and deploys their own automated agent on n8n.
- A reusable framework from one cohort to the next, updatable without being rebuilt.
A similar project?
Structuring an AI programme at your institution?
Whether it's a full curriculum or a single module to insert into an existing programme, the method starts from the target skills at graduation and works back to the entry point.