Interview with Tencent Cloud Vice President Wu Yongjian: The "Productivity Battle" of AI Agents Has Begun
Author | Huang Yu
Editor | Zhang Xiaoling
In the past three years, large model capabilities have evolved rapidly, and the global AI competition is about to enter its second half.
An emerging industry consensus is that the war of Chatbots is basically over, and the next focus of competition is shifting to "AI Agents that can get things done."
It is not difficult to see that since last year, the concept of AI Agent has rapidly gained popularity, with the imagination for applications continually expanding. Against this background, how to enable AI Agents with production-level capabilities has become a shared direction for industry exploration.
Recently, Wu Yongjian, Vice President of Tencent Cloud and head of Tencent Cloud Intelligent Product Development, pointed out in a deep conversation with Wallstreetcn that some Agent products aimed at personal scenarios have validated the potential of intelligent agents in complex reasoning and execution, showing how Agents can "think and act like humans." Tencent Cloud ADP, on the other hand, addresses the question of "how Agents can be truly implemented in complex enterprise B-end environments to generate productivity."
Focusing on deepening platform capabilities, enriching content ecology, and strengthening upper-layer applications, Tencent Cloud ADP (Tencent Cloud Agent Development Platform) has recently completed a new round of upgrades. The important task for the coming year is to enable Agents to truly become a productivity infrastructure that is reliable, governable, and sustainably evolving for enterprises through systematic evolution of engines, platforms, infrastructure, and ecosystem.
In Wu Yongjian’s view, Agents are evolving from mere tools to unified entrances for applications and services. However, this trend will be even more profound on the B-end, and will not simply copy the path of the C-end.
"Enterprises naturally possess multiple roles, systems, permissions, and compliance requirements. This means that the Super Agent for the B-end cannot be a singular intelligent entity, but must be a platform-level capability. The core is not the scale of the model, but rather the scheduling, coordination, and governance of multiple Agents," Wu Yongjian said.
Therefore, he believes, the requirements for Agents in enterprise scenarios differ from those in personal ones, leading to a structure where the cloud serves as the foundation and the Agent platform as the hub.
Tencent Cloud, anticipating the accelerating commercialization of AI Agents, fully upgraded its large model knowledge engine to Tencent Cloud ADP in May last year.
Reportedly, in the past year, Tencent Cloud ADP released six major versions and thousands of feature requests. Currently, the platform has been implemented in more than 20 industries, including finance, media, retail, and healthcare, and has opened capabilities to ecosystem partners, with the number of partners growing more than threefold in a year.
Wu Yongjian stated that the core focus of Tencent Cloud ADP over the past year can be summed up as: around "landable, scalable, operable," continuously solidifying the platform-level foundation of intelligent agents. Technically, the primary investment has been in RAG, workflow, and Multi-Agent engines to ensure agents "can run and run stably" in complex enterprise scenarios.
Regarding models, which are highly watched, Tencent Cloud ADP is not strictly bound to any particular model approach. "We offer a unified model marketplace, integrating Hunyuan, Youtu capabilities, and supporting third-party models such as DeepSeek, Zhipu, and Moonshot, while also allowing direct access to privately fine-tuned models on TI-One."
Wu Yongjian noted that for enterprises, models are replaceable resources, not platform-locked abilities. This is crucial for long-term cost control and continuous optimization of results.
If technology and the platform solve whether AI Agents "can be used," then the ecosystem determines whether AI Agents "can be used at scale."
One of the key strategies for Tencent Cloud ADP in ecosystem construction is to create benchmark application solutions with ecosystem partners. To this end, the platform itself will deepen its linkage with Tencent Cloud CVM, TKE, Lighthouse, and other IaaS products, providing partners with an integrated sales and delivery method.
In addition, another important strategy for enriching Tencent Cloud’s ecosystem is the introduction of "CB linkage."
"CB linkage": On one hand, Agent capabilities are refined in high-frequency C-end scenarios such as QQ Browser and IMA, and then systematically productized and delivered to ADP; on the other, the engineering capabilities of the B-end in stability, governance, and multi-Agent collaboration in turn enhance the C-end experience.
Wu Yongjian said this creates a positive cycle of "continuous verification in real scenarios → continuous evolution of platform capabilities," making ADP’s capabilities not just limited to labs or demos, but already validated by large-scale real users.
Reportedly, Tencent will also launch an Agent Center on QQ Browser. Agents developed via Tencent Yuanqi (the agent development platform for C-end) can be launched in QQ Browser’s Agent Center.
"If the trial shows that the Agent is active among users, it can be listed on ADP for sale. It’s like having a rapid C-end testing ground to iterate this process. In contrast, if relying solely on ADP’s B-end scenario, the validation and conversion cycle is much longer," Wu Yongjian said.
Regarding the current situation of the AI race, Wu Yongjian believes that the Agent market is moving from the early stage of "concept display, new capability experience" into a deeper stage where stable productivity delivery is the core. Agents that only "know how to chat" are no longer sufficient for real business needs; the focus is shifting rapidly to agents that are "landable and scalable."
"There are two breakthroughs in the next 1-2 years: first, reflection capability – the Agent must know how to review when it makes mistakes; second, self-evolution – the more it’s used, the smarter it gets."
In other words, if in 2025, people are still debating whether Agents can write code or book flights (C-end scenarios), the main theme of the next two years will certainly be "ROI and reliability."
Thus, Tencent Cloud’s strategy is clear: not just to make a tool, but to build a "stable productivity" platform. Relying on the connectivity of Tencent products, let Agents truly enter the enterprise business flow, making AI’s return on investment (ROI) visible and tangible.
Obviously, what will be truly scarce next is production-grade Agents that can be embedded in core business processes, bring measurable ROI, and operate stably over the long term.
Against this backdrop, the main battleground for AI Agent competition is shifting from model capability to platform capability, engineering capability, and ecosystem capability. Whoever can solve complex scenarios, scalable operations, and long-term governance issues can become the trusted "intelligent foundation" for enterprises.
Risk Warning and DisclaimerThe market involves risks, investment should be cautious. This article does not constitute personal investment advice, nor does it take into account individual user’s special investment goals, financial situation, or needs. Users should consider whether any views, opinions, or conclusions in this article are suited to their individual circumstances. Investing accordingly is at your own risk.
