Quark Achieves Over 10 Million Volunteer Reports, Setting a Domestic Record in Large-Scale Deep Research Applications}

On July 1st, Alibaba's AI flagship application Quark announced that it has generated over 10 million professional volunteer reports for students and parents. These reports are produced by the newly launched "Volunteer Report" Agent, which utilizes deep research technology with capabilities of "task planning—execution—checking—reflection." Industry experts believe this achievement has become the largest-scale application of deep research technology in China.

"Volunteer Report" is a core feature introduced by Quark this year for college application planning, and it is the only agent in China capable of generating complete volunteer reports. The agent is highly personalized and strategic, capable of planning thinking paths based on user information, utilizing search tools for information gathering, and operating the volunteer form through dedicated tools, with continuous checks and reflections during the process. The underlying support comes from Quark’s large college entrance exam volunteer model and a professional knowledge base, enabling decision-making at nearly expert levels.

During execution, the agent employs a multi-round mechanism of "tool invocation + reflection adjustment" to dynamically optimize volunteer plans. For example, when a student states "prefer to stay within the province (prioritizing Guangzhou and Shenzhen), and not consider remote areas," the system automatically generates search commands: prioritize universities in Guangzhou and Shenzhen, then consider other cities in Guangdong, and exclude remote options. If high-quality options are limited, the agent can proactively reflect and expand the search scope, such as including developed cities around Guangdong or relaxing regional restrictions for top-tier universities (like 985), giving students a better chance to succeed.

Moreover, the agent can handle complex or even contradictory demands. When faced with common conflicts like "poor math scores but wanting to apply for computer science," the system triggers a "demand clarification" process, prompting users about potential capability mismatches. This expert-like thinking path and strategic adjustment are difficult for traditional tools to achieve and are key to the true intelligent decision-making enabled by Quark’s deep research technology.

Volunteer reports exemplify the application of Quark’s deep research technology in specialized scenarios. For general deep research needs, Quark has opened invitation testing to all users, supporting structured outline generation, problem clarification paths, AI tool collaboration, and multimodal output. Among the first invited users, report downloads and active sharing rates exceeded 40%, reflecting high product recognition.

"The volunteer report is an important practice of our deep research technology," said Jiang Guanjun, head of Quark’s algorithm team. "In many professional fields, user demands are highly complex and cannot be met by conventional search or simple summaries. From the application of volunteer reports, it’s clear that deep research has truly made AI a near-expert assistant for users."