By Su Da
Tencent Cloud has unveiled what it describes as China’s first cross-platform enterprise agent portal, aiming to help companies centrally manage and orchestrate their growing fleets of artificial intelligence (AI) agents.
The product, known as the ADP Agent Portal, is designed to consolidate AI agents developed across multiple platforms into a single interface, while improving their efficiency through intelligent routing and greater visibility of how the system operates. The move reflects a broader shift among early adopters of enterprise AI, who are grappling with how to manage an expanding array of digital tools that silo information and force humans to hunt for the right tool for the right task, dragging down the very efficiency AI was promised to enhance.
According to Tencent Cloud, the portal acts as a centralize command center, allowing businesses to manage scattered internal agents in much the same way they would oversee human teams, effectively treating them as “digital employees”. The aim is to ensure agents collaborate effectively and deliver measurable value, it said.
“The future of enterprise AI does not lie in how many agents you have, but in whether they can work together efficiently and translate into real-world value,” the head of Tencent Cloud’s ADP product line said. The portal is intended to help companies not only develop agents but also manage and deploy them in a structured and sustainable way.
Bringing everything together
A central feature of the platform is its unified entry point and natural language interface. As companies have rolled out AI agents, many have found them dispersed across links, chatbots and web pages, forcing employees to navigate fragmented access points to complete tasks.
The ADP Agent Portal addresses this by allowing users to issue requests in plain language via a single search bar. The system then interprets the user’s intent and routes the query to the appropriate agent in a process the company likens to using a search engine.
The portal supports agents built not only on Tencent’s own ADP platform but also on open-source frameworks such as Dify, as well as those connected directly to large language models including Tencent’s proprietary AI model Hunyuan, DeepSeek, Kimi and Zhipu.ai’s General Language Model (GLM) series. This cross-platform compatibility enables companies to bring together previously siloed tools.
At the core of this functionality is a two-stage intent-routing mechanism. The system first performs a coarse-grained classification of the user’s request, before refining the match among candidate agents. Tencent Cloud says this reduces the risk of misclassification. The entire routing process is fully observable, allowing administrators to review how each request was handled and optimise performance.
Improving efficiency and managing costs
Beyond routing, the company says the portal addresses concerns of the real-time operational status and cost of agents — areas that have often been opaque in AI deployments. To date, Agent operations have largely remained a “black box,” where the relationship between API calls, token consumption, and actual output is murky.
The portal includes a comprehensive monitoring system that tracks everything from high-level metrics down to individual requests. Managers can track real time data on usage volume, success rates, response times and token consumption, while technical teams can drill down into specific interactions to identify bottlenecks.
In one internal pilot, Tencent used the portal’s tracking capabilities to discover that response delays in a customer service Agent were caused by inefficiencies in the knowledge base query stage. By refining query paths and invocation methods, they were able to significantly reduce overall response time.
The platform also incorporates security and compliance features. It is compatible with existing corporate account systems and permission structures, and supports private cloud deployment to ensure sensitive data remain within a client’s own environment. This is intended to address increasingly stringent requirements for data sovereignty and compliance in both domestic and international markets.
As the agentic era of AI matures, Tencent Cloud’s move suggests that the competitive frontier is shifting. It is no longer just about building the smartest agent; it is about building the most manageable ecosystem.
Source:
Caijing Tuya
