Conversation with Wu Yunsheng, Vice President of Tencent Cloud: Why Enterprise-Dedicated Agents Still Matter

```

Author | Huang Yu

As the key to winning in AI competition has shifted from “model capability” to “engineering application capability,” Agents have also replaced large AI models as the main focus of this year's World Artificial Intelligence Conference (WAIC).

At WAIC, Tencent gave a comprehensive showcase of its Agent Matrix and launched the overseas version of Tencent Cloud’s enterprise-grade intelligent agent development platform ADP 4.0.

An unavoidable question is: given that personal agents like WorkBuddy have already been adopted by many, why do enterprises still go through the trouble to build their own Agent?

In a conversation with Wallstreetcn and other media, Tencent Cloud Vice President Wu Yunsheng not only explained the fundamental reason for the existence of dedicated Agents but also pointed out several hurdles that enterprises still need to overcome in building intelligent agents.

Wu Yunsheng's answer points to one word: control.

He contrasted two approaches. If employees use WorkBuddy locally to create efficient sales processes or connect to certain systems, managers might not be able to see this or let colleagues reuse these tools. But if employees build a "sales assistant" on ADP, it can be published with one click, 30 employees can use it directly, management can see usage and permissions clearly, and it can even be embedded in the company's existing CRM or OA systems.

In other words, for personal and team productivity, using end-user desktop agents is convenient. But if agents are to be used in enterprise business and operational scenarios, there are higher requirements for the construction of agents, emphasizing integration and control. In this case, building "cloud-integrated" agents based on ADP is more suitable.

Of course, enterprises also face challenges when building their own agents.

Wu Yunsheng mentioned that a major difficulty enterprises face in deploying agents is how to connect with existing legacy systems. Much data and many processes are stored in these old systems, and legacy systems in companies that previously did not digitalize well often lack mature APIs or MCP interfaces for access.

"Legacy systems will also make changes prompted by current trends, because old systems were originally designed for people: the interfaces, interaction, and output were all human-oriented; in the future, many system design philosophies must consider design for Agents," Wu Yunsheng stated. The philosophy behind systems designed for agents is different from those designed for humans.

As Agent Loop capabilities improve, tasks previously limited by retrieval and workflow capabilities are now becoming feasible.

A closely watched question is: will Agents erode existing SaaS?

Wu Yunsheng’s answer is “no.” In his view, Agents are more of an added layer on top of existing systems to connect and orchestrate rather than to overturn and replace. The real added value is providing interfaces for decades-old legacy systems to be accessed by Agents.

Aiming to meet enterprise needs for building Agents, in May 2025 Tencent will upgrade its large model knowledge engine into the Tencent Cloud Intelligent Agent Development Platform (ADP).

According to Wu Yunsheng, after more than a year of development and the emergence of new technologies this year, clients’ acceptance of Agents has noticeably increased, and demand for Tencent Cloud ADP has also at least doubled year-on-year from last year.

It is reported that Tencent Cloud ADP is already implemented in over 30 industries, covering business scenarios such as intelligent customer service, knowledge management, and media production.

Focusing on "deployability, scalability, and operability," Tencent Cloud ADP continues to strengthen its capabilities. The latest upgraded ADP 4.0 is positioned as an AgentOps platform, providing enterprises with one-stop services for cloud Agent construction, distribution, and management.

ADP 4.0 has realized bi-directional interaction between Agents and Workflows, balancing open intentions and rigorous processes. In addition, ADP offers module capabilities through OpenAPI, including spaces, applications, skills, plugins, knowledge bases, documents, and conversations to help enterprises embed Agent capabilities into their existing workflows.

Wu Yunsheng also stated that for enterprise-grade agents, the competition is not about who can build faster, but who can run more stably and securely in actual business scenarios, creating a sustained value-generating operating system. Capability and knowledge are both essential.

Wu Yunsheng believes the application of Tencent Lexiang's Agentic knowledge base is also important: "ADP solves how to build, run, and manage agents, while the knowledge base solves what agents understand, trust, and use."

Risk Warning and DisclaimerThe market carries risks, and investments should be made with caution. This article does not constitute personal investment advice, nor does it consider the unique investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article fit their specific circumstances. Any investment made based on this is at your own risk. ```