Singapore companies expect stronger returns from artificial intelligence, but only 2% of surveyed businesses said they were fully prepared for agentic AI, according to a study released by SAP on 4 August. The gap puts data quality, workforce skills and governance ahead of technology enthusiasm as the immediate operating issue.

What the Singapore sample reported

The SAP Value of AI Report 2026 surveyed 2,600 business leaders in 13 countries, including 200 in Singapore, with research conducted by Oxford Economics. In the Singapore sample, 89% said agentic AI had moderate to very high potential to transform their organisation. SAP

Respondents estimated that total AI return on investment would rise from 19% this year to 36% in two years. Those figures are expectations reported to the survey, not audited returns, and they do not show that every company uses the same cost base or measurement method.

Why governance changes the business case

The readiness findings are more concrete for operators. SAP says 82% of Singapore respondents reported challenges with incomplete data, 27% lacked a human-in-the-loop process for agentic workflows, and 30% lacked permission and access controls for agents.

What the survey cannot prove

For Singapore’s enterprise-technology market, that gap matters because agents can act across workflows and data systems rather than only produce text. Companies evaluating them need an inventory of agents, defined access rights, reliable data and clear human intervention points before projected returns can be treated as an operating plan.

The study is commissioned and published by SAP, which sells enterprise software and AI products. Its results describe respondent perceptions and expectations; they do not independently prove that SAP technology produced the reported returns or readiness levels.

Source note

SEA Connect based this brief on SAP’s official Singapore release, including its disclosed sample size. All return, readiness and governance figures are attributed to the survey and are not presented as audited performance.