Singapore’s AI Dreams Are Crashing Into a Data Reality Check

Singapore’s AI-driven leaders are not sleeping easily at night. Nine out of ten say their entire data strategy needs to be rebuilt from scratch before AI can actually work — not tweaked or optimized but completely overhauled.

A new Salesforce study surveying 200 Singapore decision-makers reveals what many suspected, but few were willing to admit: the emperor has no (data) clothes. While 54% of business leaders confidently describe their organizations as “data-driven,” 63% of their technical counterparts say they’re actually struggling to support basic business priorities with data.

The problem isn’t a lack of ambition among Singapore leaders. AI capabilities have rocketed from the #10 priority in 2023 to #1 this year among Singapore respondents. Rather, the issue is that organizations are building AI skyscrapers on sand foundations.

Garbage in, garbage out — enterprise edition

Singapore leaders estimate 27% of their organizational data is simply untrustworthy. The predictable result? Some 84% of leaders with AI in production report experiencing inaccurate or misleading outputs.

Even worse, 66% of those training their own models admit they’ve burned significant resources fine-tuning AI with bad data. They are teaching a student with corrupted textbooks and wondering why he or she keeps failing the exam.

“To truly get the most value and context from AI models, you’ve got to get your data right,” Marc Benioff, Salesforce’s chief executive officer, warned at Dreamforce. “You have to get to more integrated solutions. You have to get the priorities right. You have to get the governance right.”

The 21% problem

But even high-quality data is useless when it’s locked away. Singapore leaders estimate 21% of their company’s data is siloed or inaccessible. It’s trapped across an average of 897 enterprise applications, with only 29% of them talking to each other.

Yet, 78% believe their most valuable business insights lie within the inaccessible 21%. It’s like knowing there’s gold in your basement, but you’ve misplaced the key.

The consequences ripple outward: reduced AI capabilities, obscured customer views, diminished personalization, missed revenue opportunities. Over 70% of Singapore’s data leaders cite all of the above.

It’s back to the fundamentals

Smart organizations are returning to fundamentals. 58% are adopting zero-copy data integration, a technique that accesses data across multiple databases simultaneously without moving or reformatting it. The results are dramatic: these companies are 74% more likely to deliver superior customer experiences and 182% more likely to succeed with AI.

“Singapore organizations facing mounting pressure to expand their AI capabilities must first get their data foundation in order,” said Gavin Barfield, Salesforce’s vice president and chief technology officer for solutions in ASEAN. “Unifying disparate data and building robust governance will be critical to unlocking real business value from AI.”

There’s also growing interest in natural language interfaces. 97% of business leaders say they’d perform better asking data questions in plain English rather than wrestling with SQL.

The key takeaway: The agentic AI revolution everyone’s chasing won’t happen until the unglamorous work of data unification, governance, and integration gets done. Sometimes you need to fix the plumbing before you can install the smart home system.

Image credit: iStockphoto/PeopleImages