The best use cases aren’t “make a course with AI”. They’re focused applications where AI helps designers analyse, generate options, adapt content and improve learning. Learning designers still need to check the output against source material, protect confidential information and ask whether the proposed activity reflects the decisions people face at work. 

1. Use AI to investigate what is getting in the way of performance

Escalations are rising. The customer-support manager asks for a course on handling difficult customers.

Before designing it, you need to understand what is going wrong.

AI can help you sift through interview notes, survey comments, support tickets and call records to identify recurring obstacles. Examples I’ve come across include a refund policy nobody can explain consistently, a CRM that makes customer history difficult to find, or uncertainty about what to say when a caller becomes angry. Each finding points to a different response.

If advisers know the policy but struggle to explain it under pressure, practice and feedback may help. If two policy documents give different answers, those documents need fixing. If the CRM hides essential information, a course will leave the obstacle in place.

The designer checks AI’s suggested themes against the source material, investigates conflicting accounts and discusses the findings with advisers and subject matter experts.

AI can organise the evidence. It’s the designer who decides what it supports and what still needs investigating.

Try this prompt: “Review these anonymised interview notes and support tickets. Group recurring performance obstacles and quote the evidence for each. Separate what people explicitly reported from your interpretation. Flag conflicting accounts and missing information. Suggest whether each obstacle may involve skills, processes, systems or motivation. Do not assume training is the answer.”

2. AI-Powered Task Analysis for Learning Design

AI can help turn an expert’s notes or explanation into a draft task analysis: the observable steps, decisions and common mistakes involved in a task such as handling a conversation about missed deadlines. The designer and subject matter expert then check the draft against real workplace practice and policy, adding the judgement and exceptions AI may miss. Reviewing a draft is often easier for a busy expert than explaining every step from scratch.

Try this prompt: “Using these SME notes, break this task into observable actions and decision points. Identify what the person needs to notice, do and check at each stage. Flag missing information and label any suggested steps that are not supported by the notes. Suggest possible mistakes for the SME to review.”

3. AI Can Support Learning Architecture and Storyboarding

AI can help turn a validated task analysis into a draft learning pathway, proposing sequences, activities and storyboards for practice. For example, it might suggest a pre-workshop scenario, live practice, workplace application and reflection for new managers learning to hold difficult conversations.

4. Generate Realistic Learning Scenarios and Practice Activities with AI

AI can help designers create realistic practice from approved product information and common customer objections. It can draft varied scenarios and role-plays that ask representatives to listen, make decisions and respond. For example, when a customer claims a competitor offers the same treatment for less. Learners can rehearse privately, repeat difficult exchanges and receive specific feedback on what they did well or missed. Designers and subject matter experts must test the scenarios and feedback for accuracy, plausibility and safety, then assess progress through the behaviours practised and evidence that they carry into real customer conversations.

Try this prompt: “Using this approved product information and these anonymised customer objections, draft three sales practice scenarios. For each, include the customer’s situation, opening objection, information they will reveal if asked, and plausible responses to the representative’s choices. Create a feedback guide tied to observable behaviours. Flag missing information and do not invent product capabilities, prices or competitor claims.”. 

5. Use AI to support learning at the moment of need 

Three weeks after a leadership workshop, a manager is preparing for a performance conversation. They remember practising it. They cannot quite remember how they started.

AI can help bring the relevant guidance back into reach.

Using the programme’s approved materials, an assistant could offer a short preparation checklist:

• What needs to change?
• What specific examples will you discuss?
• What do you need to understand from the employee?
• What will you agree before the conversation ends?

Afterwards, it could prompt the manager to reflect: “What did you learn about the employee’s situation? What did you agree? When will you follow up?”

The same approach can support other tasks. A frontline adviser could retrieve the checklist for an unfamiliar request. A sales manager reviewing a proposal could check whether it addresses the priorities the client raised.

AI can also support spaced practice. A week after a workshop, a manager might revisit a short scenario, choose a response and receive feedback linked to the approach they practised.

The designer decides when a prompt will help, which materials the assistant can draw on and when it should direct someone to a colleague. If the answer is missing from the approved guidance, the assistant should make that clear.

“Remember to use active listening” gives a busy manager little to work with, but “Before suggesting a solution, ask what has made the deadline difficult to meet” gives them something to try in the conversation ahead.

Try this prompt: “Using these approved programme materials, draft three short prompts for a manager preparing for a performance conversation, two reflection questions for afterwards and one follow-up practice scenario. Link each to a specific behaviour taught in the programme. Keep them brief and flag anything the materials do not cover.”

6. Use AI to make learning content easier to access and use

A learner may face barriers such as unexplained technical terms, colour-dependent diagrams or inaccurate captions. AI can help designers draft plainer explanations, glossaries, transcripts, diagram descriptions and translations for review. The designer checks that each version preserves the necessary information. For example, describing not just a flowchart’s colours, but the decision it requires. AI can also flag jargon, long sentences and instructions that rely on visual or audio information alone; feedback from learners with relevant access needs confirms whether the changes work in practice.

Try this prompt: “Review this learning content for unexplained terms, complex sentences and instructions that rely on visual or audio information alone. Suggest clearer wording while preserving requirements and exceptions. Draft a glossary and text alternatives for the diagrams, using their stated learning purpose. Flag anything that needs specialist or learner review.”

7. Use AI to draft assessments and feedback, and review what learning achieved

A learner can define personal data. Can they decide what to do when a colleague asks them to send a customer spreadsheet to a personal email address?

AI can help designers draft questions around decisions like this, using approved policies and realistic workplace situations.

It can suggest answer options that reflect mistakes people might make: sending the file because the colleague is trusted, removing customer names but leaving other identifying details, or checking the approved route for sharing it.

The designer and subject matter expert then check the question. Is there enough information to choose an answer? Does it reflect the organisation’s policy? Could someone pass simply by picking the longest or most cautious option?

AI can also draft feedback that explains the consequence of a choice. “Incorrect” offers little help. “Removing names may still leave customers identifiable from the remaining details” gives the learner a reason to reconsider.

For written responses, AI can help draft a marking guide with observable criteria. An assessor can then check whether it distinguishes a complete, accurate answer from one that misses a critical step.

Afterwards, AI can group learner comments and flag recurring problems. If several people describe a scenario as confusing, the designer can investigate the wording, instructions and feedback.

Those comments show where the learning experience may need attention. To assess whether performance improved, designers also need evidence from work, such as how staff handle data-sharing requests after the programme.

Where assessment affects certification, progression or someone’s job, a qualified assessor should review AI-assisted scoring and feedback before results are issued. They need to check the reasoning, resolve ambiguous answers and ensure the criteria have been applied consistently.

Try this prompt: “Using this approved policy and performance goal, draft three scenario-based assessment questions. Include plausible incorrect options, explain the reasoning behind each answer and link it to the relevant policy section. Flag ambiguity or missing information. Draft a marking guide based on observable decisions and actions for subject matter expert review.”

8. Use AI to design the questions that demonstrate impact

AI can also help designers draft the questions that demonstrate impact: what people now do differently, what barriers remain and whether measures such as errors, escalations or rework have changed. The designer ensures these questions are specific, ethical and linked to evidence the organisation can realistically collect.

Try this prompt: “Using this performance problem, learning goal and intended workplace behaviours, draft evaluation questions for learners, line managers and business stakeholders. Include a mix of behavioural, outcome and barrier questions. For each question, state what evidence it will provide, when it should be asked, and what would count as a meaningful change. Flag questions that may be unreliable, intrusive or impossible to answer with available data.”

• AI is good at processing inputs, generating first drafts, creating variations and supporting scale.
• Learning designers are good at defining the problem, understanding context, validating accuracy, protecting learners and making design decisions.

The risk is that organisations may accept plausible-looking output without asking whether it addresses a real need, reflects accurate practice or treats learners fairly.

Used well, AI gives learning designers more time for the work that matters: understanding performance, designing meaningful practice, and asking the questions that show whether learning changed what people do - and what difference it made.

The goal must be better learning design, with AI doing more of the heavy lifting behind the scenes.

Which AI use case has made the biggest difference in your learning design work, and where do you think AI should never make the final decision?

NOTE: Before trialling any of these ideas, check your organisation’s AI policies and official guidance, including requirements for data protection, confidentiality and approved tools. Obtain any necessary internal approvals before uploading information or testing an application. AI-generated outputs should always be reviewed for accuracy, suitability and alignment with organisational policy before use.