Vietnamese News Pioneer VnEconomy Charts 26-Year Path from HTML to In-House AI

Vietnam’s official connection to the global internet in November 1997 marked the beginning of a new era, though initial online applications were largely confined to email and a few websites from major technology firms. In the media landscape, a small cohort of news organizations began exploring online publishing between 1999 and 2000. Among them was VnEconomy, the online edition of Thoi bao Kinh te Viet Nam, which launched in 1998 and distinguished itself as one of the few digital publications operating as a genuine online newspaper with daily content updates.

Over the subsequent 26 years, digital publishing technology has advanced in ways that were once unimaginable. Its position as an early adopter provided VnEconomy with invaluable experience and fostered a commitment to integrating the most advanced technologies available to journalism.

Initial Experiments

The technological underpinnings of early online journalism were starkly different from today’s standards, a reflection not only of Vietnam’s nascent tech infrastructure but also of the infancy of digital publishing globally.

Sophisticated content management systems (CMS) and database-driven platforms had not yet been developed. Tools now considered essential, such as Google Analytics for measuring readership or cache servers and content delivery networks (CDNs) for performance, did not exist. Theories on digital page design were primarily adaptations of print layouts, as serious research into reader eye-tracking and on-screen interaction patterns would only emerge years later.

In its formative years, VnEconomy operated using the limited technologies of the time. The workflow for online reporters and editors was highly manual. Daily tasks involved receiving content from the print edition, converting fonts, editing static HTML files with software like FrontPage or Dreamweaver, manually inserting hyperlinks, updating the homepage, and uploading files to a server.

Repurposing content from other publications, mostly print newspapers, was even more labor-intensive. The process often required retyping articles by hand or, using a more “advanced” method, scanning printed pages, running them through optical character recognition software, proofreading the text, and then integrating it into the HTML publishing workflow. Original reporting followed a similarly manual path from writing to publication.

Several technical challenges compounded these difficulties. The Vietnamese font ecosystem was fragmented, with northern publications generally using the TCVN3 (ABC) standard while the VNI standard was prevalent in the south. Unicode would not see widespread adoption across Vietnamese websites until after 2005. Browser compatibility was another significant hurdle, as Internet Explorer and Netscape often rendered websites differently, especially concerning their support for JavaScript, which powered features like animated headlines and image effects.

Internet access in Vietnam relied heavily on slow dial-up connections, limiting the user base. Consequently, page-loading speed was the paramount concern for online newsrooms, frequently forcing them to reduce image quality to a minimum. The widespread adoption of ADSL broadband technology in 2004-2005 created the conditions for the rapid growth of online newspapers. This period also saw the increasing prevalence of CMS technology and the emergence of Google Analytics as a critical tool for monitoring audience behavior.

From Hyperlinking to AI

As one of the first news organizations to launch an online edition, Tap chi Kinh te Viet Nam, formerly Thoi bao Kinh te Viet Nam, has navigated multiple waves of technological change. The rapid development and adoption of artificial intelligence since late 2022 has catalyzed a profound transformation at VnEconomy, arguably a more significant leap than the earlier shift from static HTML pages to dynamic, database-driven systems.

Beginning in early 2023, the magazine invested in a partnership with technology company Actable AI to develop its own proprietary small language model. This raised a strategic question: why undertake the difficult task of building a proprietary model when large language models like ChatGPT and Gemini were already available and delivering impressive results?

The decision was rooted in a distinct journalistic philosophy. First, the core values of journalism are authority and accuracy, which can only be fully guaranteed when a newsroom controls the data used to train its AI systems. Second, a proprietary model can be continuously updated with new knowledge from the publication’s extensive content archives and becomes more cost-efficient over time. Third, owning the model allows for the deep integration of AI capabilities across all newsroom functions, from story ideation and topic recommendation to research and data analysis.

The magazine’s AI strategy was designed for comprehensive integration into the content production and distribution process, rather than for isolated experiments or individual use cases.

Today, AI is embedded throughout VnEconomy’s CMS, fundamentally altering newsroom workflows. Reporters can confidently use AI to assist with article production, as the system’s data sources are fully verified, eliminating the risk of fabricated information. Journalists have instant access to relevant data within the CMS while working on stories. The technology can automatically transcribe conferences, seminars, and interviews, and even generate preliminary article drafts to support reporters.

This comprehensive integration has reduced labor-intensive tasks and significantly improved productivity. The English-language edition of VnEconomy, which operates with a small team, now delivers a greater volume of content with faster publication times and reliable translation quality specialized for economic and business reporting. The daily podcast production process also requires minimal staff, as AI automates multiple stages, from selecting articles for summarization to generating audio using proprietary voice technology.

As AI becomes more powerful and accessible, its application in journalism is becoming routine. Using AI to summarize documents, create illustrations, or generate infographics is now as common as using a camera or a notebook. However, the true value of AI is validated only when it is tightly integrated into a newsroom’s standardized workflows. The critical distinction lies in moving beyond individual convenience to embedding AI into formal editorial processes. Only then can a news organization effectively monitor, manage, and quantify its impact and productivity gains.

(*) Mr. Nguyen Hoang is a Deputy General Managing Editor of VnEconomy.