Tencent Cloud’s Agent Development Platform
At the 2025 Tencent Cloud AI Industry Application Summit, Tencent Cloud revealed a significant upgrade to its large model knowledge engine, transforming it into the Tencent Cloud Agent Development Platform (TCADP). This platform integrates Tencent Cloud’s RAG (Retrieval-Augmented Generation) technology, comprehensive Agent functionalities, and features refined through real-world deployments, with the goal of precisely meeting evolving user needs.
The unveiling of the Tencent Cloud Agent Development Platform underscores Tencent Cloud’s commitment to empowering enterprise clients with the resources needed to rapidly prototype and deploy Agent-based applications. This platform serves as a cornerstone in Tencent’s broader AI strategy, enabling businesses to leverage AI Agents to enhance efficiency, automate processes, and deliver innovative solutions.
Tang Daosheng, a Senior Executive Vice President at Tencent and the CEO of the Cloud and Smart Industry Group, emphasized the enhanced capabilities now available to users, stating that they can now enable Agents to independently deconstruct complex tasks, devise execution strategies, and selectively utilize available tools. He highlighted a crucial achievement: “We have achieved zero-code support for multi-Agent handover collaboration for the first time, further lowering the threshold for building Agents.” This development significantly lowers the barrier to entry for businesses looking to integrate AI Agents into their workflows.
Within the Tencent Cloud Agent Development Platform, Tencent Cloud has curated a comprehensive Agent tool ecosystem, compliant with the MCP protocol and interoperable with key components of the OpenAI Agents SDK. Furthermore, it comes pre-equipped with a curated selection of high-quality plugins, both internal and external, including Tencent Location Services and other ecological MCP Servers. This robust selection of tools and services enables AI Agents to effectively engage with diverse data sources, leverage specialized functionalities, and expand the reach of their services.
These capabilities are meticulously designed to empower AI Agents to more effectively engage with various tools, access specialized data, and broaden the scope of their services. The platform’s emphasis on interoperability and a rich ecosystem allows developers to build sophisticated AI Agents that can seamlessly integrate with existing systems and workflows.
Across Tencent’s extensive suite of applications, a growing number of products are already incorporating Agent capabilities through the Tencent Cloud Agent Development Platform. These include QQ Browser, Tencent Health, Tencent Cloud Code Assistant CodeBuddy, and Tencent Qidian Marketing Cloud. This widespread adoption demonstrates Tencent’s commitment to embedding AI Agent technology across its diverse product portfolio.
Tang Daosheng cited QQ Browser as a compelling example, highlighting the recent introduction of the Agent QBot. This feature empowers users to issue task commands, which QBot then executes autonomously, managing everything from searches and browsing to downloading and analysis. This example showcases the potential of AI Agents to streamline user interactions and automate routine tasks.
Defining the AI Agent
While AI Agent products are proliferating rapidly, a universally accepted definition remains elusive within the industry. The lack of standardization has led to varying interpretations and expectations regarding the capabilities and functionalities of AI Agents.
Wu Yunsheng, who leads Tencent Cloud’s AI division and heads Tencent Youtu Lab, defines Agents from a user-centric perspective as a novel application paradigm characterized by autonomous planning and tool selection, including multi-Agent collaboration, to accomplish intricate tasks. This definition emphasizes the agent’s ability to independently plan and execute tasks, distinguishing it from traditional AI assistants.
In essence, Agents distinguish themselves from conventional AI Assistants, which require explicit prompts from users for each response. In contrast, Agents theoretically only need a single high-level instruction to autonomously map out and execute a complete solution. This autonomy is a key differentiating factor, enabling Agents to handle complex tasks with minimal user intervention. The underlying large language model is crucial for Agents to become genuinely useful, acting as a central “brain.”
The role of the large language model as the “brain” of the AI Agent is paramount. It provides the reasoning, planning, and decision-making capabilities necessary for the Agent to effectively perform its tasks. Without a robust large language model, the Agent would be unable to understand complex instructions, adapt to changing circumstances, or leverage available tools effectively.
Tencent’s Multi-Model Strategy
Tencent has unequivocally declared its commitment to a dual-track strategy: “steadfastly investing in self-developed models + openly embracing advanced open-source models.” This strategy underscores Tencent’s recognition of the importance of both internal innovation and external collaboration in the rapidly evolving field of AI. Since the beginning of the year, Tencent has been actively integrating the DeepSeek large model while simultaneously accelerating the iterative development of its in-house Hunyuan model.
The integration of both internal and external models allows Tencent to leverage the strengths of each approach. Self-developed models provide greater control and customization, while open-source models offer access to a broader community of expertise and innovation. This hybrid approach enables Tencent to accelerate its AI development and deploy cutting-edge solutions across its various products and services.
The Tencent-developed inference model Thinker (T1), specializing in complex tasks and deep reasoning, has undergone rapid iterations since its initial launch on the Yuanbao App earlier this year. This rapid iteration cycle demonstrates Tencent’s commitment to continuous improvement and its ability to quickly adapt to evolving user needs and technological advancements. Furthermore, Tencent has unveiled Hunyuan Turbo S, a new generation of fast-thinking models optimized for accelerated task processing.
Building on the TurboS foundation, Tencent has also introduced the T1-Vision visual deep reasoning model and the Hunyuan Voice end-to-end voice call model. Complementing these, a variety of multimodal models, including Hunyuan Image 2.0, Hunyuan 3D v2.5, and Hunyuan Game Visual Generation, have also been launched. These multimodal models extend the capabilities of Tencent’s AI platform, enabling it to process and understand information from a variety of sources, including images, video, and audio. The development and deployment of these specialized models showcase Tencent’s ambition to create a comprehensive AI ecosystem capable of addressing a wide range of user needs.
Organizational Restructuring
To facilitate rapid product innovation and deep model research and development, Tencent has integrated its AI products and applications—including Tencent Yuanbao, QQ Browser, Sogou Input Method, and ima—into the Cloud and Smart Industry Group (CSIG) this year. This consolidation streamlines decision-making processes and facilitates greater collaboration between different AI-focused teams. Concurrently, Tencent has implemented organizational changes within the Technical Engineering Group (TEG), the entity responsible for the development of Tencent’s Hunyuan large model.
Last month, sources revealed a comprehensive restructuring of Tencent’s Hunyuan large model R&D organization. Following the adjustment, TEG established two new divisions: the Large Language Model Department and the Multimodal Model Department. These entities are tasked with exploring cutting-edge technologies in large language models and multimodal large models, driving continuous iterations on foundational models, and expanding overall model capabilities. This focused approach ensures that Tencent remains at the forefront of AI research and development.
Simultaneously, Tencent is reinforcing its large model data capabilities and platform infrastructure. The Data Platform Department is focusing on the end-to-end management and construction of large model data, while the Machine Learning Platform Department is driving the creation of integrated machine learning and big data platforms. This comprehensive approach provides a robust and efficient PaaS platform underpinning both AI model training and inference, alongside big data processing, collectively supporting Tencent’s Hunyuan large model technology R&D. This investment in infrastructure is critical for enabling Tencent to train and deploy its AI models at scale.
The Agent-Driven Future
Tang Daosheng has posited that the open-sourcing of Deepseek and the breakthroughs in deep thinking signal that AI large models are surpassing the threshold of industrialization and reaching a stage of widespread deployment. This shift marks a significant milestone in the evolution of AI, signaling that the technology is becoming increasingly accessible and practical for businesses across various industries. He argues that the industry’s primary focus has shifted from model training to application and Agent-driven development.
This shift in focus from model training to application development is driven by the recognition that the true value of AI lies in its ability to solve real-world problems and enhance existing workflows. AI Agents are emerging as the key enablers of this transformation, providing a flexible and customizable platform for building AI-powered solutions.
The vast potential market for Agents is undeniably a significant factor driving Tencent Cloud’s accelerated adoption of AI Agent technologies. The ability of AI Agents to automate tasks, improve efficiency, and personalize user experiences is driving demand across a wide range of industries, creating a substantial market opportunity for companies like Tencent Cloud.
Industry Analysis and Projections
A research report from Minsheng Securities expresses a strong conviction that 2025 will be recognized as the inaugural year for AI Agents and the genesis of a software revolution. The report suggests that Agents could be a key catalyst for the revaluation of software, potentially expanding the target market for software vendors to encompass the multi-trillion-dollar labor market. AI Agents are also anticipated to enhance the consumption characteristics of software, further raising the valuation ceiling for software companies. This bullish outlook underscores the transformative potential of AI Agents and their impact on the software industry.
Gartner’s most recent forecasts indicate a substantial increase in the integration of autonomous AI within enterprise software, projecting a leap from less than 1% in 2024 to 33% by 2028. This rapid adoption rate highlights the growing confidence in AI and its ability to automate complex tasks within enterprise environments. Concurrently, over 15% of daily work decisions are expected to be autonomously executed by AI agents. In this global AI competition, AI Agents are emerging as a non-negotiable strategic imperative, leading to a broad consensus that internet giants must concentrate on both the C-end and B-end markets.
Ying Ying, Chief Analyst of Computers at CITIC Securities, highlights the contrasting approaches to Agent deployment observed across different regions. North American cloud vendors are primarily focused on facilitating efficient model and Agent deployment for their customers, while B-end vendors are more oriented towards creating and managing Agent platforms. Domestic internet giants, however, are adhering to the user traffic acquisition strategies of the internet era, aiming to capture users through general Agent products similar to “Manus,” mirroring the practices of their B-end counterparts in North America. This analysis reveals the diverse strategies being employed by different players in the AI Agent market.
Tencent’s C-End Strategy
On the C-end product front, Tencent has yet to launch a native Agent product comparable to “Manus.” While Tencent has been active in developing AI-powered features across its existing products, it has not yet released a standalone AI Agent product specifically targeted at individual consumers.
At Tencent’s recent first-quarter earnings meeting, management articulated their perspective on Agent products, categorizing them into two distinct types: general Agents that individuals can create to act on their behalf in the external world, and AI agents embedded within the WeChat ecosystem, operating within WeChat’s unique framework. This distinction highlights Tencent’s focus on both general-purpose AI Agents and platform-specific Agents that leverage the unique capabilities of its ecosystem.
Sources indicate that Tencent is building its general AI Agent capabilities through AI-native products like Yuanbao and IMA. These products serve as incubators for new AI Agent technologies and features, allowing Tencent to iterate and refine its offerings before launching a dedicated Agent product.
Tencent’s strategy entails a phased rollout of capabilities. Initially, the Agents will be equipped to provide rapid answers to questions. Subsequently, they will incorporate “chain thinking” long reasoning models to handle more complex inquiries. Over time, they will evolve to execute more complex tasks, gradually integrating “embodied intelligence” capabilities, enabling seamless interaction with other applications, programs, and even external APIs to provide comprehensive user assistance. This phased approach allows Tencent to gradually introduce new features and functionalities, ensuring a smooth and user-friendly experience.
Tencent’s management emphasizes that this is an ongoing evolution, and its capabilities are fundamentally aligned with those of general AI Agents developed by its competitors. This statement suggests that Tencent is closely monitoring the competitive landscape and is committed to developing AI Agents that are on par with the best in the industry.
The WeChat Ecosystem Advantage
The AI Agent that Tencent intends to build within the WeChat ecosystem represents a uniquely differentiated product, difficult for other vendors to replicate. The sheer scale and integration of the WeChat ecosystem provide Tencent with a significant competitive advantage in the AI Agent space.
This Agent will be deeply integrated with the core elements of the WeChat ecosystem, including social relationship networks, communication and community features, content platforms like public accounts and video accounts, and millions of mini-programs. These components collectively provide information, transaction processing, and operational capabilities spanning numerous vertical domains. This deep integration allows Tencent’s AI Agent to leverage the vast amount of data and functionality available within the WeChat ecosystem to provide a personalized and seamless user experience.
Like the previously launched native AI applications, the strategic importance of internet giants developing AI Agents lies in vying for dominance within the emerging super traffic ecosystem of the AI era, leaving no room for complacency. The AI Agent market is poised to become the next battleground for internet giants, as they compete to build the most compelling and comprehensive AI-powered solutions for both consumers and businesses.
As of 2025, the dominant theme in the AI landscape has moved from large language models to AI Agents. The focus has shifted from simply training powerful language models to building practical applications that leverage these models to solve real-world problems. The proliferation of AI Agents is inevitable, but current product capabilities remain in their infancy. The AI Agent market is still in its early stages, offering ample opportunity for innovation and disruption. In this dynamic environment, success is likely to favor those who can create the “Deepseek of the AI Agent field,” positioning themselves as leaders in the next phase of AI evolution. The race is on to build the next generation of AI Agents that will transform the way we interact with technology and the world around us.