Singapore’s latest conversation about artificial intelligence is becoming more serious — and more practical.
It is no longer just about whether people should learn to use ChatGPT, Claude, Gemini or Copilot. The discussion has moved into national AI literacy, workforce readiness, AI testing, legal accountability, cybersecurity risks and the way organisations should redesign work.
The Straits Times recently reported that Singapore wants citizens and workers to build the confidence and judgment to use AI well, with AI skills possibly becoming as basic as knowing how to use a smartphone. The same report noted plans to train about 100,000 tech-fluent workers across sectors by 2029.
IMDA has also announced that it will expand TeSA to upskill 40,000 tech professionals over the next three years under the National AI Impact Programme. The new AIxTech programme will train professionals in areas such as AI-assisted software engineering, responsible AI skills and agentic systems.
These are important moves. But for many SMEs, the real challenge is not access to AI. It is not even basic AI literacy.
The real challenge is workflow.
AI Adoption Is Not the Same as Transformation
Many businesses will make the mistake of treating AI as another software subscription.
They will buy tools, attend workshops, ask staff to “use AI”, and expect productivity to improve. Some gains will happen. Emails may be written faster. Reports may be summarised more quickly. Presentations may look more polished. Customer replies may be drafted in minutes instead of hours.
But that is not transformation.
That is task-level improvement.
The deeper productivity gains will come only when companies redesign how work is done from end to end. A recent Straits Times opinion piece made a similar point: AI adoption is not mainly a technology problem, but a leadership one. It argued that many companies are applying AI to isolated tasks instead of redesigning workflows, which is where the real gains are likely to come from.
Why This Matters More for SMEs
This is especially relevant for SMEs.
Large companies may have transformation teams, data officers, legal support, IT governance and process owners. SMEs often do not.
Many smaller companies are still founder-dependent. Knowledge sits inside people’s heads. Sales follow-ups are inconsistent. Quotation formats vary. Customer records are scattered across email, WhatsApp, Excel sheets and personal devices. Content is created only when someone remembers to do it. Reporting is reactive. Training is informal.
AI does not automatically solve this.
In fact, AI may expose the weakness more clearly.
A company with unclear processes will get unclear AI outputs. A company with poor documentation will struggle to train staff to use AI consistently. A company with messy data will not suddenly become intelligent because it has access to a powerful model.
A company with weak leadership will simply add AI noise to existing confusion.
Prompt-Writing Is Only the Surface Layer
This is why “AI literacy” must not be reduced to prompt-writing.
Prompting is useful, but it is only the surface layer. The more important skill is knowing which part of the business should change.
Should AI support sales qualification? Customer service? Inventory planning? Proposal writing? Compliance monitoring? Staff onboarding? Content production? Management reporting?
The question is not: “Which AI tool should we use?”
The better question is: “Which workflow is currently too slow, too expensive, too inconsistent or too dependent on one person?”
That is where AI can create real value.
AI Should Start With Business Pain, Not Tool Excitement
An SME does not need AI because it wants to look modern.
It needs AI if its sales team takes too long to respond to enquiries. It needs AI if management cannot see which leads are warm, which quotations are pending and which clients need follow-up. It needs AI if junior staff keep repeating the same administrative work.
It needs AI if the founder is still personally reviewing every proposal, every post, every email and every decision.
In that situation, AI is not a magic assistant.
It is a workflow redesign tool.
But redesign requires leadership.
The business owner must decide what good work looks like. The manager must define the standard. The team must agree on the process. Someone must decide what AI can draft, what humans must review, what data can be used, and what must remain confidential.
AI Safety Is Now Part of Business Discipline
This is where the current AI safety conversation matters.
The Straits Times recently reported that IMDA advised users not to give OpenClaw unrestricted access to files and applications, or run it on devices containing sensitive data. The concern is that autonomous AI tools with broad access can create risks such as unpredictable actions and data leakage.
That warning should be taken seriously by SMEs.
Many smaller companies are excited by the idea of AI agents that can read emails, access files, update systems, reply to customers and automate tasks. But without proper controls, the same convenience can become a risk.
An AI agent connected to the wrong folder, email account or customer database can create operational and reputational damage.
This is why AI adoption must come with business discipline.
Access control matters. Human review matters. Clear approval steps matter. Version control matters. Data privacy matters. Responsibility matters.
Humans Still Own the Consequences
Prime Minister Lawrence Wong recently said that Singapore’s legal frameworks were not designed for a world where machines can make consequential decisions. He noted that assumptions about responsibility, liability and accountability need to be rethought.
He also stressed that questions of responsibility, fairness and justice cannot be delegated entirely to algorithms.
That principle applies not only to law and government.
It applies to companies too.
If an AI-generated quotation contains the wrong price, who is responsible? If an AI-written customer reply makes a false promise, who approves it? If an AI tool analyses confidential client data using an unsafe platform, who is accountable? If an AI-generated report leads to a poor business decision, who checks the assumptions?
AI may assist the work, but humans still own the consequences.
AI Assurance Will Become Part of Business Credibility
ingapore is also moving toward stronger AI assurance.
A new AI Tester Accreditation Programme is expected to be launched by the third quarter of 2026 to accredit companies that test AI systems and identify weaknesses before deployment.
This is a sign of where the market is heading.
AI will not remain a casual productivity toy. It will become part of procurement, compliance, risk management and business credibility.
For SMEs, this creates both opportunity and pressure.
The opportunity is clear. AI can help smaller companies do more with less. It can support lean teams, reduce repetitive work, improve response speed, strengthen content production, support training and help owners make sense of information faster.
But the pressure is also clear. SMEs cannot afford to adopt AI in a careless way. They cannot simply let every staff member use any tool, with any data, for any purpose.
That may feel agile at the beginning, but it will become messy as usage grows.
SMEs Need an AI Workflow Audit
The next phase of SME transformation should therefore focus less on AI hype and more on AI operating discipline.
Every SME should start with a simple internal audit:
Which tasks are repeated every week?
Which decisions depend too much on the founder?
Which documents are recreated from scratch too often?
Which customer interactions are inconsistent?
Which reports take too long to prepare?
Which information is trapped inside individual staff members’ heads?
These are the places where AI can help.
But before using AI, the company must first define the workflow.
What is the input? What is the expected output? Who reviews it? What data can be used? What cannot be used? What is the acceptable level of risk? When must a human make the final decision?
This is the difference between AI adoption and AI maturity.
Singapore Needs AI Leadership, Not Just AI Literacy
ingapore is right to push AI literacy.
Workers need to understand how to use these tools. Tech professionals need to upgrade. Companies need to experiment. Policymakers need to build assurance frameworks.
But SMEs need something more specific.
They need AI leadership.
They need business owners who can see beyond the tool and ask: what kind of company are we trying to become?
They need managers who can redesign work instead of simply adding another app.
They need teams that know when to trust AI, when to question it, and when to stop using it.
Weak Workflows Will Only Become Faster Messes
AI will make strong workflows stronger.
But it will not save weak ones.
If the process is messy, AI will make the mess faster. If the data is poor, AI will make poor conclusions easier to produce. If responsibility is unclear, AI will make accountability harder to trace.
The companies that benefit most from AI will not be the ones with the most subscriptions.
They will be the ones with the clearest workflows, strongest judgment and best leadership discipline.
Singapore’s AI race will not be won by prompts alone.
It will be won by organisations that know how to redesign work.
