AI productivity hinges on data businesses can trust

Businesses hunting for productivity gains from artificial intelligence (AI) may be looking too closely at the model and not enough at what they are feeding it.

As increasingly capable AI systems enter workplaces, Antoine Yang Synology regional sales manager for Southeast Asia said the quality of their answers ultimately depends on whether businesses can give them reliable, current and accessible information.

“When using AI, the AI needs the right information from every source to give the precise answer,” Yang said during Synology’s Annual Media Event 2026 in Makati City.

The problem is that corporate information rarely sits neatly in one place.

Yang offered a newsroom as an example: an interview video may be stored on a reporter’s laptop, supporting documents buried in email and a correction arriving through a messaging app. An AI assistant asked to work from only part of that information could produce an answer that is technically sophisticated but factually incomplete.

That problem is already showing up among Philippine businesses.

Synology’s ASEAN survey found 47 percent of Philippine respondents cited data quality and availability as challenges to AI adoption, while around 40 percent said their information was scattered across different environments. Nearly 40 percent also lacked visibility into how information submitted to third-party AI tools was being handled.

Yang said organizations first have to deal with duplicate, fragmented and outdated information.

“They need to clean up the data first,” he said. Businesses must then understand and monitor how information moves through their systems as employees begin feeding it into AI applications.

The issue matters because AI is increasingly being sold as a shortcut to workplace productivity.

Around 69 percent of employees covered by Synology’s research spend between two and more than 10 hours on manual document-related work, including preparing presentations, handling spreadsheets, searching for information and translation, according to the company.

Those are tasks where AI could save time, Synology Philippines country manager Claire Huang said. But automating them does little good when the underlying information is unreliable.

“AI transformation is not just about adopting an application,” Huang said. “It’s not more AI that we need. It’s the data environment or the infrastructure behind it that needs to evolve.”

Control becomes part of productivity

Trust also means knowing where corporate information goes.

More than 90 percent of Philippine respondents said AI-ready infrastructure was important to workflow efficiency and productivity, yet 70 percent identified data security and privacy as a key obstacle to adoption.

Huang warned that employees can easily upload internal documents to third-party AI services without their IT departments knowing.

That tension may partly explain why about 74 percent of Philippine respondents favored on-premises or hybrid infrastructure, compared with just 9.9 percent prioritizing public cloud.

Huang stressed that this does not make cloud computing inherently inferior. Different workloads call for different environments, she said, with hybrid setups allowing companies to determine which information can leave their premises and which should remain under tighter control.

“Hybrid is really the main thing,” Huang said. “You are able to decide which ones you want to take ownership of.”