How AI Is Helping Government Companies in Singapore Cut Costs and Improve Efficiency

Singapore's National AI Strategy 2.0 has made AI a national necessity for public-good delivery, giving government companies a clear playbook to trim costs and raise speed by scaling talent, trusted infrastructure and governance across five priority sectors (healthcare, smart cities, education, safety and logistics).

The refreshed strategy – framed as “AI for the public good” and rolled out system-wide in 2023 – pairs investments in compute and testbeds with practical tools like AI Verify and regional LLM programmes, so agencies can move from pilots to repeatable, low-risk automation that frees staff for higher-value work.

For government teams looking to upskill quickly, practical courses such as AI Essentials for Work course details (Nucamp 15-week bootcamp) teach promptcraft and workplace AI use cases to turn strategy into measurable savings and better citizen outcomes; see the National AI Strategy 2.0 overview from EDB and course details for the AI Essentials for Work bootcamp.

Program	Length	Cost (early bird)	Includes
AI Essentials for Work course syllabus (Nucamp)	15 Weeks	$3,582	Foundations, Writing AI Prompts, Job-Based Practical AI Skills

Like any technology, AI should not be a hammer in search of a nail.

Singapore public agencies are using generative AI to shave hours off routine work: SingHealth Note Buddy AI documentation system announcement, rolling out across institutions from 4 Sept 2024, transcribes and summarises doctor–patient conversations in real time (English, Mandarin, Malay and Tamil) on Synapxe's Tandem platform so clinicians can review, edit and upload cleaner notes instead of typing them, saving roughly 2–7 minutes per visit and letting staff spend more time on care; see the announcement and GovInsider's pilot coverage for rollout and consent safeguards.

Other public tools such as NUHS's RUSSELL‑GPT are reporting roughly 40% reductions in administrative time, and private entrants like Medow's scribe (adapted for Singapore and integrated with Azure) are already onboarding clinics - together these platforms automate repetitive documentation, speed referrals and free up clinicians for higher-value tasks without replacing clinical judgment.

Tool	Deployment / Reach	Reported Efficiency
SingHealth Note Buddy AI documentation system announcement	Progressive rollout from 4 Sep 2024; supports 4 languages; >2,100 users reported	~2–7 minutes saved per patient visit
RUSSELL‑GPT (NUHS)	Deployed in NUHS cluster	~40% reduction in administrative time
Medow Health AI scribe tool Singapore launch	25 clinics in first three months; localised for Singapore	Early adopters report significant time savings

“Doctors here are curious about innovation and open to using technology that improves how they work.” - Clement Tan

Singapore's LifeSG “Moments of Life” model shows how consolidated digital services plus lightweight personalisation cut friction and staff time: by bundling birth registration, Baby Bonus and preschool waitlists into one journey, a process that once took ~60 minutes online can now be completed in about 15 minutes, and the platform pulls together data and services from dozens of back-end systems so parents don't chase different agencies; read the full LifeSG case study for details on the service design and modular approach (LifeSG “Moments of Life” service design case study).

GovTech's early thinking on “anticipatory” or push-style services underscores the goal of presenting the right service at the right moment rather than waiting for citizens to search for it (GovTech anticipatory digital services article), which - when combined with strong cross-agency task forces and modular APIs - means fewer manual handoffs, faster citizen outcomes and real cost avoidance across public workflows.

Metric	Value
Back-end systems consolidated	18 systems
APIs integrated	40 APIs
Government services surfaced	400+ services
Typical birth registration time (before → after)	~60 min → ~15 min

“We are exploring things like building moments of life platforms for businesses and as well as individuals.”

Agentic and task automation are beginning to reshape corporate services in Singapore by turning tedious compliance checklists into repeatable, auditable workflows: IMDA and the Singapore Academy of Law have piloted an agentic AGM demonstrator that can scan directors' schedules to surface viable meeting slots, generate pre‑ and post‑AGM documents, circulate reminders and track signatories, while flagging steps that need human review so statutory filings to ACRA stay compliant; read the pilot overview on GovInsider article on Singapore agentic AI pilot for corporate compliance and the technical framing on IMDA artificial intelligence research and emerging technologies page.

With the Corporate Service Providers Act in force, this kind of automation promises to cut admin cycles, reduce error-prone handoffs and free corporate secretaries to focus on strategic governance and IPO‑level advisory work - imagine a tool that turns a messy calendar scramble into a one-click, compliant AGM plan that human reviewers simply verify.

“The demonstrator is a tangible example of how AI can address real‑world challenges and elevate the way corporate governance is delivered.” - Dr Ong Chen Hui, IMDA

Domain-specific AI is already changing how Singapore diagnoses and manages chronic disease: the SELENA+ deep‑learning reader can analyse retinal photos in minutes and is being used to augment the Singapore Integrated Diabetic Retinopathy Programme (SiDRP), which screens roughly 120,000–200,000 people with diabetes a year - volume that would overwhelm manual grading without automation - so clinicians can prioritise care rather than wade through scans (SELENA+ AI retinal screening overview - Synapxe, Economic analysis of SiDRP AI-enabled diabetic retinopathy screening).

Trained on nearly 500,000 retinal images, SELENA+ has matched or exceeded grader performance (AUC ~0.94–0.96) and can cut clinician workload by up to 50%, while hybrid AI+human models flag only about 23% of cases for review - translating into fewer unnecessary referrals, lower per‑patient screening costs (manual ~$77 → hybrid ~$62) and earlier treatment that preserves sight and reduces long‑term bills; imagine a system that turns a week‑long backlog into same‑day results and a fast phone call to a patient who needs treatment now.

Metric	Value
SiDRP annual screens	120,000–200,000
Images used to train SELENA+	~500,000
Clinician workload reduction (SELENA+)	Up to 50%
AUC (referable / vision‑threatening DR)	0.936 / 0.958
Cases flagged for human review (hybrid)	~23%
Cost per patient (manual → autonomous → hybrid)	$77 → $66 → $62

AI is rewiring legal work in Singapore: LawNet 4.0 now embeds the GPT‑Legal Q&A model so lawyers can ask contract‑law questions in plain language and get context‑aware, cited answers drawn from judgments, the Singapore Law Reports and statutes - a shift that turns hours of keyword sleuthing into focused analysis and faster client strategy.

Built by the Singapore Academy of Law with the Infocomm Media Development Authority, the tool complements earlier summarisation work that has already condensed lengthy court judgments (think 8,000 words to 800) into usable headnotes, and it sits alongside an agentic AGM demonstrator that automates routine corporate‑secretarial steps so compliance stays tight while administrative cycles shrink.

For government companies and in‑house legal teams this means lower research costs, quicker issue‑spotting and more capacity for advisory work - and because the model is trained on Singapore's legal corpus and tuned to say “I can't answer” outside its remit, risk controls are built into the workflow as adoption scales.

Metric	Value
LawNet users / reach	~10,000 users; >75% of Singapore lawyers
Judgment summaries produced (earlier GPT‑Legal)	15,000+
Initial legal scope	Contract law (phased rollout to family & criminal law)
Performance improvement	Response times up to 10× faster (LawNet 4.0)

“Many lawyers have dropped out of the profession because of things like long hours. So being more efficient will hopefully help keep lawyers in the profession.” - Dr Ong Chen Hui

Shared compute is a practical lever government companies in Singapore are already using to cut AI costs and accelerate projects: the National Supercomputing Centre's ASPIRE 2A offers national-scale GPU and HPC capacity (up to 10 PFlops with 352 NVIDIA A100 Tensor Core GPUs) that researchers, agencies and firms can tap instead of buying costly local clusters, and NSCC's case studies show the same infrastructure powering work from climate models to genome sequencing (NSCC ASPIRE 2A specifications, NSCC HPC case studies).

Alongside the national backbone, commercial GPU offerings and cloud platforms - such as Sustainable Metal Cloud and telco GPU-as-a-Service pilots mentioned in recent industry write-ups - let teams pick the right balance of on‑demand cloud capacity and shared supercomputer time, so projects scale without oversized capital spend or underused racks.

The upshot: cheaper model training, faster prototyping and a shared pathway for smaller agencies to run production-grade AI without owning the full datacentre stack.

ASPIRE 2A Metric	Value
Peak compute	Up to 10 PFlops
GPUs	352 NVIDIA A100 Tensor Core GPUs
Storage	25 PBytes (spinning + nearline); 10 PBytes scratch
CPU cores	105,984

"To support this strategy and further catalyse AI activities, I will invest more than $1 billion over the next five years into AI compute, talent, and industry development." - Prime Minister Lawrence Wong (Budget 2024)

Standards, testing and risk reduction are the pragmatic backbone of Singapore's AI playbook: rather than rely on vague rules, the government built AI Verify - a voluntary testing framework and toolkit that started as an MVP in May 2022 - to let organisations run technical checks and process audits inside their own environment and produce verifiable reports (the toolkit even outputs a Docker container for easy deployment) - see the clear explainer on the AI Verify testing framework from FPF and IMDA's announcement of the AI Verify Foundation and Starter Kit.

AI Verify bundles open‑source tools (SHAP, AIF360, Adversarial Robustness Toolkit, Fairlearn), maps to international standards (NIST and ISO crosswalks) and has evolved into GenAI‑specific work: a Global AI Assurance Pilot paired 16 testers with 17 companies to refine real‑world LLM tests, and IMDA's Testing Starter Kit gives teams step‑by‑step checks for hallucination, data disclosure, undesirable content and adversarial prompts.

For government companies, that means cheaper deployments and lower legal and operational risk because problems are caught in controlled testing rather than in live services - turning compliance from a cost centre into an early‑stage quality gate that saves time, money and public trust.

Initiative	Launch / Status	Purpose
AI Verify (testing framework & toolkit)	May 2022 (MVP)	Voluntary self‑assessment of AI systems; technical + process checks
AI Verify Foundation	June 2023	Open‑source community to develop testing tools, standards and best practices
Global AI Assurance Pilot	Feb 2025	Sector pilots to codify GenAI testing norms (testers ↔ companies)
Testing Starter Kit for GenAI	May–June 2025 (consultation)	Practical safety tests for LLM risks: hallucination, disclosure, adversarial prompts

“The private sector with their expertise can participate meaningfully to achieve these goals with us.” - Josephine Teo

Singapore's funding stack makes AI adoption and automation far less risky for government companies and SMEs: Enterprise Singapore's Enterprise Development Grant (EDG) will co‑fund qualifying costs (third‑party consultancy, software, equipment and internal manpower) at up to 50% for local SMEs - and for sustainability‑related projects the support is enhanced to as much as 70% from 1 Apr 2023 to 31 Mar 2026 - see the EDG details on Enterprise Singapore for eligibility and claims guidance.

For faster, pre‑scoped tool adoption the Productivity Solutions Grant helps cover approved IT solutions (typical co‑funding ~50% with caps on some solutions), while Market Readiness Assistance can underwrite overseas pilots and promotions (enhanced cap of S$100,000 per new market through Mar 2026).

Applications are routed through the Business Grants Portal and usually take 8–12 weeks, and EDG reimbursements are paid after audited deliverables - in short, a project that slashes repetitive labour can have the bulk of its technology and consultancy bill underwritten, turning a risky pilot into a financed, auditable step toward scale (Enterprise Singapore Enterprise Development Grant (EDG) eligibility and claims, GoBusiness Singapore government grants overview and application portal).

Scheme	Support	Notes
Enterprise Development Grant (EDG)	Up to 50% (SMEs); up to 70% for sustainability projects (1 Apr 2023–31 Mar 2026)	Funds consultancy, software, equipment, internal manpower; claims reimbursed after deliverables
Productivity Solutions Grant (PSG)	~Up to 50% (caps on solutions)	Pre‑scoped IT solutions and equipment; faster adoption route
Market Readiness Assistance (MRA)	Up to 50%; enhanced cap S$100,000 per new market	Covers overseas promotion, business development and market set‑up (extended to Mar 2026)
StartupSG Founder / Tech	Founder: up to S$50,000; Tech: POC/POV grants (up to S$400k / S$800k)	Early‑stage funding for startups, proof‑of‑concept and commercialisation support

Singapore is turning AI skilling into a direct cost‑cutting lever: paid, industry‑aligned pipelines shrink recruitment cycles and reduce churn by training locals in place rather than hiring at market premiums.

National programmes from AI Singapore and IMDA combine on‑the‑job routes like the award‑winning AI Apprenticeship Programme (AIAP) - with six‑ and nine‑month tracks, real‑world project phases and a $4,000 monthly training stipend - with IMDA's push to create 800 new training and job opportunities and new frontier tracks to place hundreds of AI practitioners into companies; see the AI Apprenticeship Programme official page and the IMDA ATxSG and DEB press release on GenAI and industry adoption for details.

Employer‑backed routes such as IMDA's TechSkills Accelerator Company‑Led Training (CLT) mean businesses can capture trained talent with structured mentorship instead of paying external recruiters, while apprentices come out job‑ready (AIAP placement rates are over 90%).

The result is a leaner HR bill: paid apprentices doing billable, supervised work from month one, faster onboarding and fewer costly hiring mistakes - picture a six‑month cohort turning into a steady stream of vetted AI engineers for the organisation.

Initiative	Duration / Spots	Key detail
AI Apprenticeship Programme (AIAP) official programme page	6 or 9 months; next batches (2026)	$4,000 monthly stipend; >90% graduate placement; real-world project phase
IMDA ATxSG and DEB press release on GenAI and industry adoption	~800 new training & job opportunities	Partnerships + frontier programmes to place 800 practitioners and support GenAI adoption
IMDA TeSA Company‑Led Training (CLT) programme page	Varies (up to 12 months)	On‑the‑job training, employer mentorship, aligned to Skills Framework

“Singapore's value lies not just in our capabilities, but in our consistency – in being a partner you can count on, even when the world is less certain.” - Tan Kiat How

Public–private collaboration and concentrated R&D clusters are the engine room for Singapore's AI cost and efficiency gains: coordinated efforts from EDB, IMDA and the joint office Digital Industry Singapore (DISG) turn conferences and summits into deal‑making labs - at the AsiaTech x Singapore sidelines DISG hosted a closed‑door AI Startups & VC networking where 24 high‑growth startups pitched to 12 prospective VCs - while anchor investors and global corporates (AMD, PwC, Thales, Alibaba Cloud, ST Engineering) commit local hubs, funding and technical partnerships that let government companies tap ready talent, shared compute and validated solutions before scaling.

This stacked ecosystem - designed to move pilots into repeatable production - pairs grant and compute incentives with industry co‑innovation, so agencies can prototype faster, avoid wasted spend on bespoke stacks, and contract proven vendors with local support; read the EDB round‑up for the latest ecosystem moves and see how DISG streamlines industry engagement for scalable AI partnerships.

Metric	Value
Startups pitched to VCs (DISG event)	24 startups → 12 VCs
New GenAI projects (next 12 months)	Up to 500 projects
New AI training opportunities (next 3 years)	~800 opportunities

“Our position on the fragmented regulatory landscape is that we need to build more bridges.” - Chan Ih‑Ming, Executive Vice President, EDB

Start with a tightly scoped proof‑of‑concept, not an all‑or‑nothing overhaul: Singapore's AI Singapore 100 Experiments (100E) programme - 3‑month PoC or 6‑month MVP lets government companies fast‑track value with a focused 3‑month PoC or a 6‑month MVP (co‑funding available and access to AI engineers plus AIAP apprentices), making it realistic to validate an idea in 90 days and hand over production‑ready artefacts and documentation.

Pair that delivery playbook with safety gates from IMDA's testing work - use IMDA Project Moonshot / AI Verify testing benchmarks to red‑team hallucinations, bias and data disclosure before citizens see the service.

When the PoC proves out, scale via national incentives and partners: the AI Cloud Takeoff under the Enterprise Compute Initiative (Google Cloud credits and implementation partners) bundles bootcamps, implementation partners and Google Cloud credits - up to SG$500,000 of incentives - to stand up an AI Centre of Excellence and avoid one‑off bespoke stacks.

The practical roadmap is therefore simple: assess & scope, run a funded PoC, test with AI Verify, then scale using cloud credits and CoE tooling - so savings start appearing while governance and capacity grow in lockstep.

Path	Duration	Outcome	Team / Funding
100E (3‑month POC)	3 months	Validated Proof‑of‑Concept	2–3 AI Engineers & Project Manager; lower commitment
100E (6‑month MVP)	6 months	Production‑ready solution	4–6 AI Engineers & Project Manager; co‑funded up to SGD$150,000

Project Moonshot is one of the first tools in the world to bring benchmarking, red‑teaming, and testing baselines together - so developers can focus on what's important to them.

Singapore's future savings are not hypothetical - they're baked into policy, pockets and practical programs: Budget 2025's Enterprise Compute Initiative (ECI) and other enhanced supports give agencies and government companies access to cloud credits, compute and consultancy (EDB's quarterly AI round‑up outlines these measures), while expanded SkillsFuture and TeSA tracks put hands‑on GenAI training within reach so teams can automate routine tasks without costly hiring rounds; see how AI‑driven citizen services and chatbots already cut friction in “Moments of Life” journeys and frontline support in the CMSWire case study.

Combine funded compute, pre‑scoped grants and targeted upskilling - plus tested safety tools and procurement pilots - and the pathway to measurable cost reduction is clear: fewer bespoke datacentres, faster PoCs, and staff repurposed from data‑entry to high‑impact oversight.

For teams looking to build those practical skills quickly, the AI Essentials for Work bootcamp - 15‑week workplace AI course offers workplace promptcraft and hands‑on use cases to turn policy and infrastructure into real, recurring savings.

“To support this strategy and further catalyse AI activities, I will invest more than $1 billion over the next five years into AI compute, talent, and industry development.” - Prime Minister Lawrence Wong (Budget 2024)