Text by | Haishan

Source | Bowang Finance

The AI sector has witnessed significant milestones, from the sudden rise of DeepSeek to the unexpected surge in popularity of Manus, and now, the long-awaited launch of Alibaba's QianWen. Has AI technology truly reached a new zenith? Recently, the public beta version of the Alibaba QianWen app made its debut, marking a full-fledged entry into the AI-to-consumer (C-end) market and sparking widespread market enthusiasm. Meanwhile, Kuake has fully integrated the QianWen dialogue assistant.

01

Alibaba QianWen: What Surprises Does It Hold?

Let's first take a brief look at why Alibaba embarked on developing this application.

According to The Paper, the product manager of the Alibaba QianWen project provided a detailed explanation during an interview with the tech media outlet 'LatePost.' The timing for entering the consumer market is optimal. On one hand, the model maturity of Qwen3-Max has achieved globally leading performance and effectiveness.

On the other hand, the maturity of the entire Agent ecosystem, encompassing both third-party and Alibaba Group's internal ecosystems, has reached a stage where models can be universally applied to solve a broader range of problems.

At that time, some argued that, objectively speaking, it was still in its infancy and had not truly advanced to the point of effectively solving many practical issues. Looking back, whether it's large models or other AI applications, domestic companies are all striving to make strides in addressing real-world problems.

For instance, earlier this year, Manus's AI agent application made its debut as the 'world's first general-purpose AI agent,' causing a significant stir. Its beta invitation codes were quickly resold for nearly 100,000 yuan on secondhand platforms. At the time, Manus's key selling point was its 'end-to-end autonomous execution.' However, actual user experiences revealed flaws: when faced with complex tasks (such as multi-step SEO optimization), the output reports were criticized as 'mechanical frameworks' lacking in-depth analysis of industry trends. The high hype led to a gap between early user expectations and actual capabilities, with some scenarios still requiring manual corrections to the output. Additionally, the complex interface design and steep learning curve for operation procedures deterred some users.

In contrast, as an application launched by a tech giant, Alibaba QianWen has received significantly different evaluations from industry insiders. In their view, Alibaba excels at problem-solving. According to Lei Feng Network, the QianWen app is positioning itself as a rival to overseas competitor ChatGPT, aiming for a head-to-head competition with the 'strongest AI' overseas.

Data from Qimai shows that on November 12th, the estimated number of downloads for the QianWen app was 12,700, and by November 16th, this figure had surged to approximately 34,800. It currently remains at the forefront of the App Store's free chart. Through the QianWen app, it can be observed that its interface layout is broadly similar to other native AI applications, featuring an interactive area at the bottom. Users can input their needs through both voice and text, as well as upload various types of materials such as images, documents, and recordings. Similar to other models, QianWen also offers core functionalities commonly found in similar products, such as AI-generated images, intelligent writing, photo-based problem explanation, and translation.

02

A More Far-Reaching AI Strategy

From an industry perspective, many users hope that AI applications will adhere to a logic of 'autonomous execution of complex tasks → delivery of complete results.' The core of this closed-loop capability lies in a multi-agent architecture and cloud-based asynchronous processing, which differs somewhat from early AI large models. Most AI language models operate by searching, organizing, learning, reasoning, and generating question-and-answer responses. In contrast, complex applications can elevate simple Q&A dialogues to the next level of post-dialogue actions, achieving a new height of human-machine collaboration.

At the technical level, the entire industry is characterized by a situation where 'everyone possesses similar capabilities.' However, returning to the public beta version of the QianWen app, it can be considered another crucial step in Alibaba's AI strategy implementation after the company announced a 380 billion yuan investment in AI infrastructure earlier this year.

Public reports indicate that Alibaba plans to integrate full-scenario services such as maps, food delivery, ticketing, office work, learning, shopping, and health into the app, enhancing its 'problem-solving capabilities' and clearly targeting two core goals: 'comprehensiveness' and 'ease of use.'

In terms of benchmarking against competitors, ChatGPT is currently the global benchmark for being 'comprehensive and easy to use.' Alibaba's straightforward display of confidence in benchmarking against international first-tier enterprises is akin to creating a 'comprehensive AI gateway' for China in the domestic market.

Domestically, there are AI players such as Baidu, Tencent, ByteDance, Huawei, and DeepSeek. With the participation of numerous players, domestic large models have captured a significant market share through open-source, free strategies and performance comparable to international mainstream models. However, leading AI applications suffer from severe homogenization, and the industry is still mired in early-stage chaos and intense competition. The ultimate winner remains undetermined.

Nevertheless, Alibaba QianWen does possess certain advantages at its own level.

For example, the latest 'China GenAI Market Insight: Panoramic Research on Enterprise-Level Large Model Adoption, 2025' released by international market research firm Frost & Sullivan shows that in the first half of 2025, the total daily average consumption of large models in China's enterprise-level market was 10.2 trillion Tokens, with Alibaba's Tongyi model accounting for 17.7%, ranking first. The performance and coverage of Alibaba's foundational models can, to some extent, bring value.

03

Accelerating the Commercialization of the AI Industry

Zooming out, this event holds significant implications.

Previously, QuestMobile's latest report indicated that in August 2025, native AI apps, in-app AI, and AI assistants from smartphone manufacturers continued to grow steadily, with user bases reaching 645 million and 529 million, respectively. In contrast, the PC side underperformed expectations, with both web-based and client-based AI applications experiencing an 8.8% decline in user numbers month-over-month.

This implies that in the AI era, smartphone apps remain an irreplaceable core user gateway, while wearable devices such as smart glasses and watches have yet to gain significant traction. This is precisely the core logic behind some major players' recent focus on launching new mobile-end products. Only by rapidly acquiring users and securing a market share advantage can they establish a solid foundation and seize the initiative in the fiercely competitive AI landscape.

Whether AI application products are targeted at consumers (C-end) or businesses (B-end), opinions on innovation vary among individuals. The public is more concerned with the underlying logic and core technologies, especially when it comes to a Chinese company. Most people have become accustomed to comparing such 'groundbreaking' technologies with those of previous or overseas enterprises. Without careful consideration, the first team to 'eat the crab' (a Chinese idiom meaning to be the first to try something new and potentially risky) may face either widespread criticism or adulation.

Objectively speaking, Alibaba QianWen's role in promoting the entire AI industry cannot be overlooked, especially in terms of scenario implementation and ecological integration. More importantly, as various companies innovate in AI applications or other fields, it will ultimately benefit the further development of AI Agent intelligent agents (referring to 'intelligent entities' capable of perceiving their environment, making decisions, and executing tasks).

This viewpoint is not without merit. The coming years are expected to be the first year of agent commercialization. Survey data from Deloitte indicates that by 2025, 25% of enterprises using generative AI will deploy AI Agents, with this figure growing to 50% by 2027. IDC similarly predicts that by 2028, at least 15% of daily work decisions will be autonomously made by Agentic AI, and 33% of enterprise software applications will incorporate Agentic AI.

Against this backdrop, numerous tech giants, including NVIDIA, Baidu, OpenAI, ByteDance, and Honor, have already launched various application entities based on their own technologies. For instance, Zhao Ming, the former CEO of Honor, claimed at the launch of the Honor Magic7 series that they had created the world's first operating system equipped with intelligent agents and derived the YOYO intelligent agent, even ordering 2,000 cups of coffee for the audience on site.

This implies that amidst the development of various AI software applications, intelligent agents have significant potential for growth and explosion. Manus's execution effects further demonstrate that higher-level AI Agents are not limited to generative AI dialogues and responses. They can mobilize end-to-end resources to fully understand various forms of human information, such as images, voice, and text, and form autonomous decisions in complex situations. Under an Agent collaboration framework, they can also collaborate with other intelligent agents (multimodal agents). For example, Manus's current practice of directly assisting people with stock and real estate investments reflects this to some extent.

This precisely confirms that the accelerated commercialization of AI technology may prove effective, especially in the paradigm shift from 'suggestion generation' to 'task execution.' For instance, in finance, healthcare, and manufacturing, intelligent agents can assist humans with many low-frequency work tasks (such as report generation and contract review), freeing up human resources for creative endeavors. They also hold rich application scenario value in other fields such as culture, tourism, education, and healthcare.

The boundaries of the artificial intelligence industry are constantly being broken. From the innovation and optimization of DeepSeek's underlying algorithms and training models to the launch of Alibaba QianWen, it is evident that there is no single, unified definition of AI innovation. Industry progress is made through exploration.