Tencent Executive Urges Workflow-Focused AI Alongside Model Development

Dowson Tong, Tencent’s senior executive vice president and president of the Cloud and Smart Industries Group, is advocating AI built around specific business workflows while the company continues developing foundational models.

Tencent Cloud republished an essay tied to the company’s internal annual publication Zhidiǎn on Aug. 27. The essay calls for applying models to defined tasks, integrating them into product and engineering work, and building service agents for individuals and businesses. It also backs continued investment in foundational models.

The recommendations align with a position Tong stated in January: “Customers want choices. Different use cases need different models.” The August essay applies that idea to product development, pairing model choice with tools and services designed for particular tasks.

Tencent introduced WorkBuddy, creative tool Miora and TokenHub for global markets on May 28. The company described TokenHub as a gateway to its own models and leading third-party models. The announcement brought the three products together as part of Tencent’s global-market AI offering.

Tong said in a June 5 interview that WorkBuddy and CodeBuddy remained in an investment phase and that the Buddy team had no commercialization targets.

Tencent’s March investor presentation forecast more than doubling investment in new AI products in 2026 compared with 2025. The company’s definition excludes AI work for existing products and services, along with GPU purchases for external Tencent Cloud use. The forecast therefore covers a specific new-product category rather than all of Tencent’s AI work.

Tencent has also described ongoing model development. In a Sept. 16, 2025, announcement, the company said its Hunyuan team had released more than 30 models over the prior year. Tencent also said CodeBuddy reduced coding time by 40% and increased research and development efficiency by 16%.