Qwen's Jiang Guanjun: Personal Agent requires the synergy of three elements: model, memory, and ecosystem

Related topics top start Related topics top end Futures promotion begin Futures promotion end Miaopai begin Miaopai end Video player start Video player end Market chart begin Market chart end

Sina Tech News, October 23 afternoon: At the 2026 China Computer Congress, Jiang Guanjun, head of AI algorithms at Alibaba's Qwen business unit, shared his view that "the true Personal Agent stage should bring about the creation of incremental value and personalized proactive services; models, memory, and ecosystem are the three elements that jointly build this core capability."

Jiang Guanjun divides the development of AI applications into three stages: conversation, agent, and assistant. Currently, AI applications are accelerating toward the assistant stage, where, on the basis of delivering specific tasks, they begin to combine professional experience and personal context to autonomously plan and execute around user goals, and further discover needs that users have not yet noticed or have overlooked.

He stated that the theoretical upper limit for Personal Agents in solving problems in daily work, study, and life is close to 100%, and the key to unlocking this potential lies in better matching system capabilities with problem-solving paths, and finding optimal solutions for different problems through efficient coordination of model, memory, and ecosystem capabilities.

At the model level, Qwen places greater emphasis on the stability of task execution. To this end, Qwen has built a user demand overview covering more than 30 leading industries, identified more than 8,000 atomic model capability requirements, and introduced industry experts to participate in model evaluation and optimization, continuously improving the model's ability to understand real user needs and complete complex tasks.

At the memory level, Jiang Guanjun stated that a Personal Agent needs to understand users over the long term, identify the connection between historical information and current needs, and avoid the noise and inference costs brought by simply accumulating conversations. For example, AI can combine the pension situation mentioned in an elderly person's historical conversations to provide advice on their child's employment issue that better fits the family's circumstances. Qwen designs its memory system by analyzing event correlations and stores and retrieves memories in layers according to the strength of associations, improving the precision of memory use.

At the ecosystem level, Jiang Guanjun believes that with the continued development of model capabilities and Harness technology, the ecosystem's connection and collaboration capabilities will become key to the development of Personal Agents. Qwen is building a model-centric open service system that integrates professional knowledge, tools, industry Skills, and third-party Agents, connects services such as e-commerce and local life, as well as professional capabilities in fields such as healthcare and education, to improve the execution of complex real-world tasks.

The coordination of the three elements ultimately points to proactive services: memory provides personal background, the model combines professional knowledge to identify potential needs and plan actions, and the ecosystem supports execution. For example, in health management, it continuously pays attention to rehabilitation arrangements and their impact on daily life, and in the financial management process, it reminds users to pay attention to new risks and opportunities.

The latest news shows that Qwen's personal dedicated AI assistant "Qwen Go!" has launched a new feature, "My Apps." Users do not need programming; they can easily build applications for scenarios such as life, work, and study simply by stating their needs. The relevant feature is currently in invitation-only testing. news_keyword_pub,stock, iframe below body iframe below body General statement General statement QR code at end start QR code at end start

QR code at end start QR code at end start Editor name and work code

Editor name and work code