Workplace Learning in the AI Era: Build Capability Where Work Actually Happens

AI is changing the way work gets done. But for many Singapore enterprises, the response to an AI skills gap still starts and ends with one familiar solution: send employees to a course.

The logic is understandable. Employees need to learn how to use AI tools, so the company provides training. People attend the session, complete the exercises, receive a certificate, and return to their usual jobs.

Then, very often, not much changes.

The tools may be available, but employees are unsure when to use them. Managers may encourage experimentation but have no clear way to assess whether the output is reliable. Existing workflows may still require the same manual steps. Performance measures may reward speed without considering quality or verification. And employees who are already managing a full workload may struggle to find time to practise what they learned. This creates an important distinction between AI training and AI capability.

The World Economic Forum's Future of Jobs Report 2025 estimates that 39% of workers' existing skill sets will be transformed or become outdated between 2025 and 2030. That means organisations cannot treat capability building as a one-off training event. As technology and workflows continue to change, employees need opportunities to learn, practise, receive feedback and apply new skills directly in their roles.

This is where workplace learning becomes particularly important. Instead of taking employees away from work to learn about AI in isolation, organisations can make the workplace itself part of the learning environment. The starting point is not the latest AI tool. It is the work that needs to improve.

Start With the Work, Not the AI Tool

One of the easiest mistakes organisations can make is choosing a technology first and then looking for ways to use it.

For example, a company may introduce an AI assistant because it wants employees to "use AI more". Employees are then shown how to write prompts, generate summaries or analyse information. But what business problem is the AI actually solving?

A more useful approach starts with the workflow.

Perhaps the sales team takes too long to prepare quotations. Customer service employees spend too much time drafting repetitive responses. Managers receive operational reports but struggle to identify the issues that require attention. Technicians spend considerable time searching through maintenance records before diagnosing recurring equipment problems.

These are learning opportunities because they involve real tasks, decisions and performance outcomes.

From Course Completion to Performance Improvement

Course completion is easy to measure. An organisation can report how many employees attended a workshop, how many completed an online module and how many certificates were issued.

But these numbers do not necessarily tell the organisation whether capability has improved. Consider two employees who both complete the same four-hour AI course. One returns to work and occasionally uses an AI chatbot to rewrite emails.

The other uses AI to prepare a first draft of a customer response, checks the answer against the company's approved policy, identifies information that needs human confirmation and escalates unusual cases to a supervisor.

Both employees completed the course. But the second employee has developed a much more meaningful workplace capability. This is why learning outcomes should be connected to observable work behaviour.

Managers can assess whether employees are using the tool appropriately, checking the right information and maintaining the required quality standard. The question becomes less about whether employees know a particular feature and more about whether they can perform the task effectively.

Build Learning Around Short, Practical Sprints

Workplace learning does not have to mean months of classroom training. For many AI-related capabilities, a short learning sprint can be more useful because employees repeatedly apply what they learn.

A simple structure could look like this:

1. Demonstrate

An experienced practitioner demonstrates how the task can be performed using AI. The demonstration should use a realistic business case rather than a generic example.

2. Practise

Employees practise using historical or low-risk cases. This gives them room to make mistakes without immediately affecting customers, operations or business decisions.

3. Review

A supervisor, experienced colleague or coach reviews the work.  The feedback should focus not only on the final output, but also on how the employee reached it.

4. Apply

Employees use the approach on appropriate live work. At this stage, the organisation can begin measuring whether the new way of working actually improves performance.

5. Reinforce

The organisation updates examples, checklists, standard operating procedures or team practices based on what employees learn. This final step is often overlooked. If the training ends but the workplace remains unchanged, employees can easily return to old habits.

The Skill That Matters Most: Knowing When to Verify

Much of the conversation around AI learning focuses on prompting. Employees are taught how to write better instructions, ask follow-up questions and structure prompts.

These skills can be useful, but they are only part of the picture. In many workplaces, the more important capability is verification.

AI can produce an answer that looks convincing while still being incomplete, inaccurate or unsuitable for the specific situation. Employees therefore need to understand what should be checked before an output is used.

That might include:

• Checking information against an approved company source

• Confirming figures against the original dataset

• Reviewing whether an answer complies with internal policies

• Protecting confidential or personal information

• Testing assumptions before using an AI-generated recommendation

• Identifying situations that require human judgement

• Escalating high-risk or unusual cases

The verification process will look different depending on the role. A customer service officer may need to check whether an AI-generated response accurately reflects the company's current policy and maintains the right tone. An HR analyst may need to verify calculations, challenge assumptions and ensure that personal information is handled appropriately. A technician may need to compare AI-generated troubleshooting suggestions against approved maintenance procedures and actual machine readings.

In each case, the employee is not simply learning how to get an answer from AI. They are learning how to judge whether the answer is good enough to use. That is a much more valuable workplace capability.

Make Good Judgement Visible

Experienced employees often perform verification almost automatically. They know which information looks suspicious. They know which source to check. They know when a situation falls outside the normal process.

The problem is that this knowledge is often tacit. A less experienced employee may not know what an experienced colleague is looking for. Workplace learning can make that judgement visible. One practical method is to create annotated examples.

Managers Are Part of the Learning System

Even well-designed training can struggle if managers are not involved. Employees may learn a new way of working during a workshop but return to a team where the manager still expects the old process.

For example, an employee may learn how to use AI to produce a first draft more efficiently. But if the team's performance target focuses only on the number of tasks completed, employees may feel pressure to prioritise speed over verification.

Managers therefore have an important role in turning training into workplace behaviour. They do not necessarily need to become AI experts.

They do, however, need to understand:

• What the employee is expected to do differently

• What good performance looks like

• Which parts of the process require human judgement

• What should be checked before an AI output is used

• When an employee should escalate an issue

• How the new workflow will be measured

Give Employees Time to Practise

There is another practical issue that organisations sometimes underestimate: learning requires time. Asking employees to "learn AI" on top of an already full workload may sound efficient, but it can result in superficial experimentation. Employees may attend a course, try the tool once or twice and then abandon it when work becomes busy.

A better approach is to deliberately allocate time for practice. That does not necessarily mean taking employees away from their jobs for an extended period. A learning sprint could include a protected hour each week to practise on selected cases, followed by a short review with a supervisor or peer.

The important point is that practice is treated as part of the work rather than an optional activity employees have to complete after everything else is finished. This also allows organisations to identify capability gaps early. If several employees struggle with the same step, the issue may not be individual ability. The workflow, instructions or training design may need to change.

Measure What Changed, Not Just What Was Delivered

This approach also changes how organisations evaluate learning.

Traditional learning metrics might include:

• Number of participants

• Attendance rate

• Course completion

• Assessment scores

• Employee satisfaction

These measures still have value, but they should not be the end of the conversation. For workplace learning, organisations should also consider measures connected to the original business problem.

If the objective was faster quotation preparation, measure turnaround time. If the problem was customer response quality, review accuracy and rework. If the goal was reducing repetitive administrative work, measure time saved while checking that quality has not deteriorated. If the goal was improving maintenance diagnosis, examine troubleshooting time, repeat faults and adherence to approved procedures. The exact metrics will depend on the role and business objective.

Workplace Learning Does Not Mean Abandoning Formal Training

Formal training still has an important role. Employees need a solid understanding of AI basics, data protection, cybersecurity, responsible use and the limitations of different tools. But training alone is not enough. The real value comes from giving employees the chance to apply that knowledge to the work they actually do, with guidance and feedback along the way.

For many organisations, the practical approach is to combine a common AI fundamentals programme with role-specific learning sprints. Everyone gets the same baseline, but teams then learn how AI applies to their own work, whether that means analysing data in finance, handling customer queries or supporting operations. The technology may be similar across the business, but the skills, risks and judgement required can be very different.

Workplace Learning Does Not Mean Abandoning Formal Training

Formal training still has an important role. Employees need a solid understanding of AI basics, data protection, cybersecurity, responsible use and the limitations of different tools. But training alone is not enough. The real value comes from giving employees the chance to apply that knowledge to the work they actually do, with guidance and feedback along the way.

For many organisations, the practical approach is to combine a common AI fundamentals programme with role-specific learning sprints. Everyone gets the same baseline, but teams then learn how AI applies to their own work, whether that means analysing data in finance, handling customer queries or supporting operations. The technology may be similar across the business, but the skills, risks and judgement required can be very different.

Build Capability Where Work Actually Happens

At Cadence, we believe workplace learning should go beyond simply putting employees through a course. The focus is on helping organisations build practical capability around the skills, workflows and challenges that matter to their business. By connecting learning to real work, employees can put new skills into practice while organisations can see how that learning translates into better performance.

Whether you are looking to build AI capability, strengthen workplace skills or develop more effective learning programmes, Cadence can help you turn learning into something employees can apply—not just something they complete.