How to Identify AI Training Needs in a Singapore SME

Why start with tasks rather than AI tools?#

Starting with tasks keeps training tied to real work. If you begin with a tool, staff learn features they may never use. If you begin with a task, you can see what good output looks like, which skills are missing and what could go wrong, then choose training that closes that specific gap.

SWDA's article on job redesign for SMEs makes a similar point. It advises investing in AI "across workflows and processes, not just at the level of specific job roles". It also says organisations need to consider data security, data governance and digital infrastructure. As the adviser quoted in the article puts it: "You don't redesign a role for the sake of it."

How do you list the tasks worth training for?#

Run a short task review with managers and the staff who do the work. Ask where people repeatedly rewrite information, search internal documents or prepare similar summaries. For each task, note how often it happens, what a good output looks like, any current quality problems and what kind of information it uses.

1. Pick one team or workflow, not the whole company.
2. Ask each person to name two or three tasks they repeat every week or find difficult.
3. For each task, record how often it happens and what a good result looks like.
4. Note existing quality problems, such as errors, delays or rework.
5. Mark the type of information involved: public, internal, client or personal data.
6. Set aside tasks that involve sensitive information until you understand your privacy and security controls.

SWDA's job redesign article also points to roles with high turnover, roles that fall short of industry benchmarks and roles easily made more efficient with technology, while keeping a "human touch".

How do you build an SME training-needs matrix?#

A training-needs matrix puts each task on one row, with the role, the possible AI use, the skill gap, the risk level and a priority. It turns a long list of ideas into a short, ranked plan. The table below is an illustrative template, not data: replace every row with tasks from your own review.

How to fill in each column#

• Skill gap: describe what the person cannot do yet, for example "cannot tell whether a summary left out a key point".
• Risk level: base it on the data involved and the cost of a wrong output. Personal data, client data and anything sent to customers rate higher.
• Priority: favour frequent, lower-risk tasks with clear quality problems. High-risk tasks come after controls are in place.

Worked example: scoring one row#

Take the customer service row. Replies go out every day, so the task is frequent. A wrong price or promise reaches a customer, so the risk is medium. The skill gap is specific: staff need to give the AI approved product notes and check every reply before sending. That makes it a strong candidate for practical training in prompting and evaluation.

Which staff need AI awareness and which need building skills?#

Not every employee needs the same course. Most staff need to understand what AI can and cannot do. Some need stronger prompting and evaluation skills for daily tasks. A smaller group may need to build and test simple AI applications. Match each person to the level their row in the matrix actually calls for.

• Awareness: capabilities, limits and company data rules. Suits most staff and managers.
• Prompting and evaluation: giving context and checking outputs. Suits staff who use AI daily.
• Application building: planning, building, testing and launching a simple AI app. Suits people who will create tools for colleagues.

GenAI Builder Sprint sits in the third group. It is a 2-day classroom course where adults plan, build, test and launch a simple AI app. No previous AI or programming experience is required, but applicants must meet the formal entry requirement of at least a Level 3 professional qualification, an Advanced Certificate or equivalent.

A 2-day application course is not necessarily right for every employee. Our post on how to choose AI training for employees compares the main types of training.

How do you check readiness, privacy and practical constraints?#

Before nominating anyone, check that learners meet the course entry requirements and have the confidence to take part. Then check the practical side: devices, approved AI tools, language and time away from work. Finally, set clear rules on what company information staff may put into external AI services.

SWDA's job redesign article describes a common SME pattern: one premium AI chatbot subscription shared across teams, where much client information cannot be shared and staff must anonymise data or rewrite prompts. Data rules need attention, not just skills.

If your company deploys its own AI system, the PDPC's Advisory Guidelines on Use of Personal Data in Generative AI say deployers must safeguard personal data, including in new data sources such as prompts and generated outputs. This includes educating users on the specific types of personal data that should be input. The guidelines also encourage clear written policies. Our post on responsible AI at work covers privacy, accuracy and human oversight.

What government support can an SME use for this work?#

Some SWDA support covers the needs analysis itself, not only the training. Under the Enterprise Workforce Transformation Package, the enhanced job redesign grant can fund consultancy that includes assessing a company's AI readiness. SWDA's OurWorkforce portal also offers employers digital tools to identify opportunities for transformation, redesign jobs and adopt AI effectively.

According to SWDA's factsheet on the Enterprise Workforce Transformation Package (EWTP):

• From March 2026, the SkillsFuture Workforce Development Grant (Job Redesign+), or WDG(JR+), provides funding support of up to 70% of workforce transformation and job redesign project costs for SMEs, capped at $150,000 per enterprise.
• Its Workforce Consultancy component, with a sub-cap of up to $50,000, may include assessing a company's AI readiness, human capital or career health needs.
• To qualify, an enterprise must be registered or incorporated and operating in Singapore, and have at least three local employees (Singapore Citizens or Permanent Residents).

For course fees, SWDA lists Enhanced Training Support for SMEs, with up to 90% course fee subsidies for a wide range of SWDA-funded courses. Check that the course, run and employee qualify before you book. Our funding overview explains the main schemes.

What does this look like in a Singapore SME?#

Consider a 12-person logistics firm in Jurong whose managers want staff to "use AI more". Instead of buying one course for everyone, the operations manager runs a short task review, fills in a training-needs matrix and finds that customer replies, delivery updates and procedure writing come up most often.

An illustrative plan might look like this:

• Everyone attends a short awareness briefing on what AI can do and the firm's rules on customer and staff data.
• Two customer-facing staff practise prompting and checking replies before sending.
• One operations executive who meets the entry requirement takes a seat in a public class of GenAI Builder Sprint, to learn how to build and test a simple internal tool such as a procedure lookup.
• HR ideas wait until the firm has written rules on personal data.

After training, each person demonstrates one skill against a simple rubric, such as spotting an error in an AI draft. Any claim about time saved waits for a comparison with a baseline taken before training.

Our corporate AI training page explains employer-sponsored enrolment. GenAI Builder Sprint runs as public classes, so companies book seats for staff in a scheduled class. This is training support, not a business consultancy.

What are the common mistakes when identifying AI training needs?#

The most common mistake is choosing the training before understanding the work. Others include sending everyone on the same course, starting with tasks that involve sensitive data, skipping readiness checks and claiming productivity gains that nobody measured. Each one wastes training time or creates risk that a simple matrix would catch early.

• Starting with a tool: staff learn features, not how to improve a task.
• One course for everyone: awareness, evaluation and app-building are different needs.
• Sensitive tasks first: wait until privacy and security controls are clear.
• Skipping readiness checks: nominees who miss the entry requirement or cannot get time away struggle to finish.
• No baseline: without a before measure, you cannot show what changed.

Sources#

1. Job Redesign in the Age of AI: How SMEs Can Level Up, Skills and Workforce Development Agency (SWDA), accessed 7 November 2026
2. Advisory Guidelines on Use of Personal Data in Generative AI, Personal Data Protection Commission (PDPC), accessed 7 November 2026
3. Factsheet on Enterprise Workforce Transformation Package (EWTP), Skills and Workforce Development Agency (SWDA), accessed 7 November 2026
4. SWDA Launches New Initiatives to Strengthen Careers, Skills Development and AI Readiness, Skills and Workforce Development Agency (SWDA), accessed 7 November 2026
5. Training Subsidies & Levy, Skills and Workforce Development Agency (SWDA), accessed 7 November 2026

Frequently asked questions#

Where should an SME start with AI training?+

With its work, not with the latest AI tool. List the tasks people repeat or find difficult, note how often they happen, what a good output looks like and what information they use, then decide which skills are missing.

What is an SME training-needs matrix?+

A simple table with one task per row and columns for the role, the possible AI use, the skill gap, the risk level and a priority. It turns a long list of ideas into a short, ranked training plan.

Does every employee need the same AI course?+

No. Most staff need to understand what AI can and cannot do, some need stronger prompting and evaluation skills, and a smaller group may need to build and test simple AI applications. A 2-day application course is not necessarily right for every employee.

What should we check before nominating staff?+

Whether learners meet the course entry requirements and have the confidence to take part, plus devices, approved AI tools, language and time away from work. Also set clear rules on what company information staff may put into external AI services.

Is there government support for assessing AI readiness?+

SWDA's factsheet on the Enterprise Workforce Transformation Package says the Workforce Consultancy component of WDG(JR+) may include assessing a company's AI readiness. Eligibility rules apply, including being registered or incorporated and operating in Singapore with at least three local employees.

How can we follow up after training?+

Ask employees to demonstrate a skill against a simple rubric, for example spotting an error in an AI draft or explaining why certain data should not be shared. Do not claim productivity gains without comparing results to a baseline.