How to Personalise AI-Generated Learning for Different Employee Roles
AI is making it easier for organisations to turn existing documents, procedures and internal expertise into learning content. But generating a course is not the same as making it relevant to the people who need to learn from it.
Employees differ in their professional experience, prior knowledge, communication preferences and responsibilities. A standardised learning experience may be too basic for some people, too complex for others, or insufficiently connected to their everyday work.
Effective AI-powered learning personalisation starts with understanding the learner, not just the content.
To personalise learning meaningfully, AI needs more than a source document and a topic. It needs a structured understanding of the intended audience and a way to use that information when generating learning content.
Why AI-generated learning needs personalisation
Traditional approaches to digital learning often start with a subject, define a learning objective and produce a course for a broad audience. AI can accelerate this process, but speed alone does not solve the challenge of making learning relevant to different people.
Consider an organisation introducing a new operational procedure. An experienced technician, a newly hired employee and a team manager may all need to understand the same procedure. However, they may have different levels of prior knowledge, different responsibilities and different needs when applying the information.
This challenge becomes particularly relevant in large organisations. Poste Italiane's training programme for 800 distribution centre directors illustrates how existing training materials can be transformed into short digital learning experiences for a large managerial audience. In a different operational context, Alperia's programme for 800 field technicians addresses the challenge of delivering technical knowledge to employees working across locations and in four languages.
These cases illustrate different corporate learning requirements. The underlying company knowledge may remain the same, while the audience, context and way that knowledge needs to be applied can differ.
AI-generated learning can address these differences, provided the system has enough information about the intended learners to make meaningful choices during content generation.
Personalisation should therefore go beyond changing the tone, shortening a lesson or simplifying its vocabulary. It should take into account the characteristics of the people learning and the context in which they will use what they learn.
What is a Learner Model?
A Learner Model is a structured representation of the characteristics relevant to an individual's learning experience or to a group of learners.
In AI-powered learning, it provides information that can guide how content is generated for a specific audience. Instead of treating the source material as the only input, the AI can also take into account who the content is intended for.
For corporate learning, a useful Learner Model needs to account for more than job title or seniority. Two employees with the same role may have different levels of experience, confidence, familiarity with a topic and preferences for how information is presented.
A multidimensional model provides a more complete basis for adapting learning content to these differences.
The challenge is to identify which dimensions are relevant, how they complement one another and how they can be translated into practical guidance for AI-generated learning.
Five dimensions for personalising AI-generated learning
A structured Learner Model can describe a target audience across five complementary dimensions. Each provides a different perspective on the people for whom the learning experience is being designed.
1. Professional Profile
Professional characteristics help establish the learner's role and position within the organisation.
Relevant variables include:
Seniority: the learner's level of responsibility or position.
Experience: their professional experience.
Function: the business function or functions to which they belong.
Decision Autonomy: the degree of autonomy associated with their work.
These characteristics help contextualise the content around the learner's responsibilities and the decisions they may need to make.
For example, content about a new operational process may need to emphasise practical execution for an employee responsible for carrying it out, while addressing coordination and decision-making for a manager.
The distinction matters when designing learning for different professional audiences. Poste Italiane's programme for distribution centre directors provides a concrete example of learning designed for a specific managerial population rather than an undifferentiated employee audience.
2. Prior Knowledge
Personalisation also depends on what learners already know about a subject.
Relevant variables include:
Knowledge Level: the learner's existing level of knowledge.
Topic Familiarity: how familiar they are with the specific subject.
Confidence: how confident they feel about their knowledge or ability.
These variables provide a basis for adapting how a topic is introduced and developed.
Learners who are new to a subject may need foundational explanations and clear examples. Those already familiar with it may benefit from a greater focus on nuances, exceptions or practical application.
Knowledge and confidence are not necessarily the same. A learner may be familiar with a topic but uncertain about applying it. A useful personalisation approach needs to leave room for these distinctions.
This is particularly relevant in compliance learning, where employees must understand rules and apply them in their work. In this pharmaceutical compliance training case study, existing training content was transformed into shorter, multilingual learning experiences for a global workforce. The case illustrates the challenge of making specialist knowledge accessible across a diverse audience, although it does not isolate the effect of individual learner characteristics on learning outcomes.
3. Communication Preferences
People can differ in how information is explained and presented.
Relevant variables include:
Explanation Style
Examples
Tone
Vocabulary
These preferences can guide the way information is communicated without changing the underlying knowledge or learning objective.
For example, a learner may benefit from concrete examples, while another may prefer a more direct explanation of a process. Vocabulary and tone can also be adapted to make content more accessible and appropriate to its intended audience.
The objective is not to assume that one communication style is universally effective for a particular type of person. It is to give the AI relevant parameters with which to shape the learning experience.
This becomes particularly important when content must reach people across different linguistic and professional contexts. Alperia's field-technician training programme, for example, addressed the delivery of technical learning in four languages. The case illustrates the importance of making content accessible to a distributed workforce, while language adaptation remains distinct from personalisation based on an individual learner profile.
4. Cognitive Profile
A Learner Model can also describe dimensions relevant to how information is presented and applied.
Relevant variables include:
Abstraction Level
Reasoning Style
Theory vs Practice
Decision Style
These dimensions can help guide the balance between conceptual explanations, concrete situations, practical exercises and decision-oriented examples.
For instance, a learning experience intended for employees who must execute a procedure may need to emphasise practical steps and realistic situations. A different audience may need more emphasis on the underlying principles or on evaluating alternative courses of action.
These dimensions are intended to guide content design. They should not be interpreted as fixed psychological classifications or as proof that learners belong to rigid learning styles.
The aim is to give AI a richer set of parameters for making content-generation decisions, rather than relying on a single label such as job title or seniority.
5. Business Context
Learning takes place within a specific organisational and operational environment.
Relevant variables include:
Industry
Business Priorities
Industry provides context for terminology, examples and operational situations. Business priorities help connect learning content to the outcomes and challenges that matter to the organisation.
For example, the same general knowledge may need to be contextualised differently for employees working in manufacturing, retail or financial services. Likewise, a company prioritising safety may need learning content to emphasise risk recognition and correct procedures.
The importance of operational context is clear in Gruppo Frassati's approach to training blue-collar employees in the flow of work, where access to learning needs to fit around operational shifts. In a different setting, Sicuritalia reduced site-transfer onboarding time by 90% by addressing the information employees needed when moving between sites.
These examples show why learning content needs to reflect the situations in which employees must apply their knowledge. They illustrate the importance of context, rather than proving the independent impact of a particular Learner Model dimension.
From a Learner Model to personalised content generation
Defining learner characteristics is only the first step. Those characteristics must also inform the way AI generates the learning experience.
A practical personalisation workflow can be understood through four steps.
1. Start with existing company knowledge.
The source material may be a procedure, a product document, an internal guide or other existing organisational knowledge. This provides the subject matter that the learning content needs to preserve and communicate.
2. Define the intended audience.
The Learner Model describes the relevant characteristics of the target cohort across professional profile, prior knowledge, communication preferences, cognitive profile and business context.
3. Use the profile to guide content generation.
The AI uses the audience information alongside the source material and learning objective to shape the resulting content. Depending on the context, this may influence explanations, examples, vocabulary, the balance between theory and practice, and the way concepts are contextualised.
4. Evaluate the resulting learning experience.
The generated content still needs to be assessed for accuracy, relevance and suitability for its intended audience. A structured Learner Model provides a basis for personalisation, but does not by itself guarantee that the result will be effective.
This approach allows organisations to work from a common source of knowledge while adapting how that knowledge is presented to different groups of employees.
An example: the same knowledge, different audiences
Imagine a company introducing a revised procedure for handling a customer issue.
The company has one authoritative source document, but several groups need to learn from it.
New employees may need clear definitions, foundational explanations and step-by-step examples.
Experienced employees may already understand the underlying process and need to focus on changes, exceptions and situations that require particular attention.
Team managers may need content that connects the procedure to operational decisions, team coordination and business priorities.
These are illustrative examples, not prescribed outcomes for particular roles. The appropriate content depends on the actual learner profile, the learning objective and the source material.
The same principle applies to knowledge-intensive subjects outside traditional operational training. Fondazione Valter Longo's case study on longevity and nutrition illustrates a different challenge: turning complex scientific knowledge into short learning experiences that people can engage with over time.
The important principle is that personalisation does not require every employee to start from a different body of company knowledge. It requires the learning experience to be designed with the intended audience in mind.
How zick learn approaches AI-powered learning personalisation
zick learn applies this principle to the transformation of existing company knowledge into learning content.
Its approach builds on research conducted with Berkeley into the dimensions needed to represent a learner comprehensively and on learning-related heuristics that informed its first AI model for transforming corporate content.
The current model uses a structured Learner Model to guide the generation of content for a defined audience. It organises learner characteristics across the five dimensions described above, giving the AI a richer representation of the people for whom the learning experience is intended.
Rather than relying solely on the source document or a broad job-role label, the model is designed to take account of multiple characteristics that can influence how learning content should be presented and contextualised.
This approach is relevant across different kinds of corporate learning, from technical procedures and compliance to managerial development and knowledge sharing. For example, Alpitour's case study on keeping travel coaches aligned across markets illustrates the challenge of maintaining consistent knowledge and communication among people working across different markets.
These cases show the variety of audiences and operational contexts that corporate learning must serve. They should not, however, be read as controlled evidence of the Learner Model's independent impact.
What to consider when evaluating AI learning personalisation
Organisations evaluating AI-powered learning solutions should look beyond whether a platform can automatically generate a course.
Several questions can help distinguish basic content generation from a more structured approach to personalisation:
What does the AI know about the intended learners? Does it use a structured learner profile, or does it rely only on job title and a general description of the audience?
Which dimensions influence the generated content? Can the system account for prior knowledge, professional context and communication preferences, as well as role?
How is personalisation applied? Does it influence the explanations, examples and contextualisation, or only the length and tone of the content?
How is content quality assessed? Are accuracy, relevance and suitability evaluated rather than assumed from the fact that AI generated the material?
Can the approach support different audiences at scale? Can the organisation adapt learning experiences across functions, seniority levels and business contexts without having to create every version manually?
These questions help clarify what a platform means when it claims to personalise AI-generated learning.
Conclusion: personalisation starts with a model of the learner
AI can make it easier to transform company knowledge into learning content. But generating content efficiently and personalising it meaningfully are two different challenges.
Meaningful personalisation requires a structured understanding of the intended audience and a way to translate that understanding into content-generation decisions.
A multidimensional Learner Model provides a framework for doing so. By accounting for professional profile, prior knowledge, communication preferences, cognitive profile and business context, AI can be guided to produce learning experiences that are more closely aligned with the people they are intended to serve.
For organisations looking to turn existing know-how into practical employee skills, the key question is not simply how AI can generate more learning content.
It is how AI can generate the right learning experience for the people who need to apply that knowledge at work.
zick learn develops this approach by combining AI-powered content transformation with a structured Learner Model designed to support deeper personalisation of corporate learning.
