Anthropic adds AI skills pact to Singapore’s training push

Singapore’s AI training opportunity lies in helping employers change daily work, a task Anthropic’s new skills pact places alongside responsible adoption.

AI company Anthropic announced the agreement alongside its Singapore office opening on 14 October. Its memorandum of understanding with the Ministry of Digital Development and Information covers cooperation on AI safety and responsible adoption. The company plans to explore support for small and medium-sized businesses and help build AI capability among public officers.

For training providers, the commercial opening is in turning those objectives into work an employer will pay to implement. A provider needs a defined buyer, a delivery role and enough teaching volume to cover course development. An exploratory pact gives it a reason to pursue a partnership; a commissioned programme gives it revenue.

Public officers already have a training route

The public-sector opportunity sits within a wider government skills programme. On 2 March, Singapore announced plans for the Institute of Digital Government to equip more than 150,000 public officers with digital, data, design and AI skills. The ministry said the institute would design and deliver curricula aligned with government priorities.

All public officers will need to complete foundational modules covering cybersecurity, data protection and AI literacy. The institute’s planned targeted courses include product thinking, data governance and AI applications. Anthropic therefore enters a market with a curriculum organiser and a defined audience, which favours providers able to fit specialist teaching into an existing structure. Digital training for Cabinet ministers and senior public service leaders had also started, Minister of State for Digital Development and Information Jasmin Lau said.

There is also an audience already using Anthropic’s tools. The company says more than 1,500 engineers and technical staff at the Government Technology Agency of Singapore, known as GovTech, have adopted Claude Code to modernise government systems and speed software delivery. That creates a practical basis for teaching around actual tasks, beyond introductory explanations of AI.

The delivery economics will depend on how agencies divide the work. Central course design can support consistent standards, while specialist providers can contribute exercises and assessment tied to particular jobs. Direct teaching by Anthropic would leave less room for an intermediary selling basic product instruction. Local providers have a stronger proposition when they connect tool use to an agency’s operating requirements.

SME courses need an operational payoff

Small businesses present a different sales problem. Singapore’s National AI Impact Programme aims over the next three years to support 10,000 local enterprises in integrating AI into business processes and 100,000 workers in becoming ‘AI bilingual’. Those targets create a broad policy-backed audience, but each employer still needs a reason to release staff and spend money on delivery.

Demand for practical instruction predates Anthropic’s agreement. CNA reported that nearly 400 leaders at Singapore University of Technology and Design were training to build custom AI solutions. Aventis Graduate School, CuriousCore and Vertical Institute had also recorded increases in demand for custom AI courses. The pact enters an active market with providers already competing for learners.

For an SME, a course built around a recurring administrative task has a clearer purchasing case than general product familiarity. Teaching can cover selecting a suitable task, checking the output and deciding which information staff can use safely. The provider’s value lies in leaving the employer with a workable process and employees who can operate it without constant assistance. PwC’s Global Workforce Hopes and Fears Survey 2026 found that 74 per cent of Singapore respondents had used AI at work in the past year, up from 69 per cent in 2025. Half said AI had increased their work complexity, while 46 per cent said it had added to their workload. Another 62 per cent said they were using more skills in their jobs. Only 65 per cent said they had access to the learning and development resources they needed, compared with 70 per cent globally.

That approach also changes course economics. Employer-specific exercises take time to prepare, and follow-up support uses instructor capacity after the classroom session ends. Providers need to price that work explicitly. A short introductory course can attract interest, but implementation support offers a clearer route to repeat business if the employer can see an operational gain.

Existing provision raises the bar for paid delivery

Technical training already has a detailed competing proposition. The Infocomm Media Development Authority and AI Singapore plan AIxTech, a programme covering AI-related software engineering and responsible use. Its first phase comprises 10 self-paced online modules lasting about 18 hours, with coding exercises and learner support. Eligible final-year Information and Digital Technologies students can attend free.

AIxTech participants will have access to several coding tools, including Claude, Codex and GitHub Copilot. Its planned content covers automating software engineering tasks, integrating AI tools into workflows and developing agentic AI systems. This gives employers a broad alternative to a course centred on one supplier and puts pressure on providers to explain the value of specialist instruction. The programme’s second phase includes post-course support through online and in-person practitioner communities and advanced follow-on modules. Responsible AI training will also use IMDA AI Verify Foundation governance frameworks and testing tools for traditional, generative and agentic AI systems.

For local operators, the largest commercial risk is spending ahead of commissioned demand. A delivery appointment with funded cohorts would justify instructor recruitment and course development. An arrangement that leaves employers to buy individually would require a separate sales effort. Providers can pursue both routes, but each needs its own cost base and revenue assumptions.

Safety teaching offers a possible area of specialisation, especially when it helps staff make decisions during routine work. It also demands more than a presentation on responsible use. Anthropic’s planned cooperation includes technical exchanges with Singapore’s AI safety institutions on testing standards and incident response. Anthropic joined the Generative AI Evaluation Sandbox run by the Infocomm Media Development Authority and AI Verify Foundation in 2023. It later worked with those organisations on red-teaming, testing designed to uncover weaknesses in AI systems, across English, Mandarin, Malay and Tamil.