HONG KONG, July 9, 2026 /PRNewswire/ -- The demo was unremarkable in its simplicity. A young product manager, five months into her job at Kingsoft Office, typed a command. The screen flickered. Customer data streamed in from a CRM system. A 300-line procurement document was parsed, matched, and evaluated. A post-project report assembled itself.

It is a small moment, easily overlooked. But it speaks to something larger — and more urgent — than a single software demonstration.

Across the corporate world, a quiet frustration is building. Companies are pouring billions into artificial intelligence. They are buying chatbots, subscribing to copilots, and running pilot after pilot. Yet in boardrooms and break rooms alike, the same complaint echoes: the AI answers questions, but it does not do the work.

A Gartner survey published in April found that more than 70% of enterprise AI projects are failing to meet expectations. Bain & Company, in a separate study of 951 companies, reported that AI investment is surging — but returns are not following. The gap between promise and delivery has become the defining anxiety of the corporate AI era.

At LEAP EAST, the international technology conference that opened in Hong Kong on July 8, Kingsoft Office, the Chinese software company behind WPS Office, offered a diagnosis and a remedy. The problem, the company argues, is not that AI is not powerful enough. It is that AI has not been allowed to touch the actual work.

"We've moved from the era of 'how to make one person faster' to 'how to make an organization smarter,'" said Xu Liu, Vice President of Kingsoft Office. "That requires a fundamentally different architecture."

The company's answer is WPS 365, an enterprise productivity platform that embeds AI directly into business workflows — not as a separate assistant, but as a layer that reads, writes, analyzes, and acts inside the documents and systems that companies already use. Kingsoft Office calls it an "Enterprise Brain Execution Layer."

The most striking part of the offering is Qingzhou AI, a lightweight private-deployment system designed for companies that cannot afford the risk — or the cost — of traditional AI infrastructure. In an era when data sovereignty has become a geopolitical and regulatory concern, Qingzhou AI runs on a single server, reducing hardware requirements by more than 95%. It is an approach that turns the prevailing industry logic on its head: instead of asking companies to build bigger infrastructure, it asks how little infrastructure is truly needed.

The implications for Asia Pacific are significant. Across Southeast Asia, small and medium enterprises make up the vast majority of businesses. Many operate in sectors — finance, legal, education — where data cannot leave national borders. For them, the choice has long been between expensive on-premise systems and risky public cloud AI. Qingzhou AI offers a third path.

Kingsoft Office has been in the document business for 38 years. That longevity, the company believes, gives it an advantage that newer AI-native startups cannot replicate: an intimate understanding of how work actually flows through documents — contracts, reports, spreadsheets, meeting notes — and how AI can live inside them, not alongside them.

"What is changing is not just the software," Liu said. "It is the very idea of where intelligence lives in an organization."