Skip to content
SME AI Impact

Aggregate analysis

What 30 verified implementations have in common

Every figure on this page is computed from the structured dataset behind the case studies. Nothing is estimated or drawn from outside the record.

These are award winners, selected for demonstrated business impact. They are not a representative sample of Singapore SMEs, and the distributions below should not be read as national adoption statistics.

Macro benchmark context

Singapore Enterprise AI Landscape

Source: IMDA Singapore Digital Economy Report 2025 & The Straits Times
Award competition scale

290 / 21

290 nominations across 21 sectors submitted for the inaugural 2026 awards; 30 selected (10.3% acceptance).

SME national AI adoption

23.4%

Up from 14.5% in 2024. Almost 1 in 4 Singapore SMEs now deploy AI tools across their daily business operations.

Productivity gain rate

87.6%

Of Singapore AI-adopting enterprises report productivity/process improvement as a primary benefit, validating the gallery's heavy time/capacity focus.

Cost reduction rate

44.0%

Of adopting enterprises achieve cost reduction or resource optimisation — contextualising operational wins like Blue Aqua's 30% feed efficiency.

Distinction note: The 87.6% productivity and 44% cost reduction figures describe all AI-adopting enterprises surveyed in IMDA's national digital economy study, providing an economy-wide reference point. The 30 award winners in this showcase were chosen from 290 nominations for verified operational tenure and quantified outcomes.

Winners by sector

Sectors as published in the award citations. Seventeen sectors appear across the thirty winners; media and creative is the largest group.

Innovation versus adoption

15 winners in each category, by design.

  • 15Innovation Excellence
  • 15Adoption Excellence

Research evidence distribution

A grade measures available evidence, not the quality of the implementation.

AI use cases by business function

Assigned from the described workflow, not from marketing language.

Verified stack types

The dominant technical approach in each implementation, as described in the record.

Published benefit types

A winner can report more than one type of benefit, so counts exceed 30.

Documented technology layers

How often each layer appears in a case-study stack table.

Named vendors in the record

Counts of case studies that name a vendor at any layer. Absence is not evidence of absence.

Research coverage distribution

Ten evidence dimensions per case study, expressed as a percentage.

Sector by AI use case

25 of 255 combinations contain a winner. The matrix is sparse, which is the finding: these companies did not converge on one sector-specific pattern.

Counts of winners by published sector and assigned AI use case.
SectorWorkflow automationComputer visionContent generationDocument processingKnowledge retrievalCustomer serviceGenerative mediaClinical decision supportDesign optimisationInventory and costingMarketing operationsOperations schedulingOrder processingPlatform engineeringPredictive analyticsTotal
Media & Creative325
Education213
Food & Beverage1113
Built Environment112
Business Services112
ICT/Consultancy112
Retail112
Transport & Mobility22
Agriculture11
Aquaculture11
Environmental Services11
Healthcare11
Logistics11
Marine and Offshore Energy11
Personal Care Services11
Talent & Recruitment11
Wholesale Trade11
Total54333221111111130

Reading the numbers

Three patterns hold across the set, and one caveat matters more than the patterns.

Documentation is the exception. 11 of 30 case studies name at least one technology vendor. 19 describe what the system does without naming what it runs on. The award citation rewards business outcomes, so the technical record is thin by design rather than by omission.

Most winners automate a queue. The recurring shape is an intake of unstructured work, an extraction step, a structured record, and a human approving exceptions. Invoices, orders, enquiries, footage and documents account for the majority of the set.

Time is the most reported benefit. Time saved appears in almost every citation, followed by cost and capacity. Revenue growth is claimed less often, and usually by companies selling the capability rather than only using it.

Caveat: the percentages and multiples in these citations are company-reported. No independent audit of the figures was published, and this showcase does not recalculate them.

Research notifications

Research updates

Get notified when evidence grades change or the 2027 edition publishes.

No marketing or sales outreach. Strictly research revisions and dataset releases.