Welcome to the world of AI agents. Also known as ‘agentic AI,’ these digital ‘coworkers’ can do everything from sourcing suppliers for a small business to creating and deploying webpages — all without prompting by human users.
A demonstration of Google’s AP2. Credit: Google
Unlike AI chatbots that provide responses to user requests, AI agents can be trained to break down complex tasks into workable steps, and then automatically complete them.
A number of high-profile AI agents have emerged this year from both established and new players. In the U.S., Google rolled out its automated shopping system Agent Payments Protocol (AP2) this fall, while Microsoft has launched its Microsoft’s “control plane” for AI agents, Agent 365.
This week, The Wire is highlighting some of the leading AI agents emerging in China, where they are as diverse as the needs they serve.
AGENTS ADVANCE
Big companies such as Alibaba and ByteDance, which have access to enormous amounts of user data from their various platforms and apps, have been able to release multiple AI agents specifically tailored to the needs of their existing users.
U.S. markets consolidate toward two or three winners with massive capital moats. China has seven or eight serious players exploring genuinely different approaches — and the pack keeps growing.
Afra Wang, a tech writer and author of the Concurrent newsletter
Meanwhile new players, such as Manus.Ai and Zhipu, have developed agents which can conduct complex research and analysis and create webpages and reports.
Experts say China’s agentic AI landscape is defined by diversity and competition.
“U.S. markets consolidate toward two or three winners with massive capital moats. China has seven or eight serious players exploring genuinely different approaches — and the pack keeps growing,” says Afra Wang, a tech writer and author of the Concurrent newsletter.
A demonstration of Alibaba’s Accio. Source: Accio
Julian Goldie, chief executive of marketing firm Goldie Agency and a frequent user of Chinese AI agents, says they often have more functionality compared to Western models, in addition to being largely free and open-source.
“The speed at which they move seems to be a lot faster,” he says. “[Alibaba’s] Qwen seems to deliver new updates every week at this point…it’s very aggressive.”
AI agents are gaining acceptance in workplaces, despite their potential to replace existing employees.
A survey of almost 3,000 employees and executives in global enterprises, conducted in May and June this year, demonstrated general acceptance of AI coworkers, but not yet managers.
However, their emergence comes with risks. In November, AI research company Anthropic alleged that a state-sponsored Chinese hacking group had utilized the company’s own agentic coding tool “to execute the cyberattacks themselves.”
Savannah Billman is a Staff Writer for The Wire China based in NYC. She previously worked at the National Committee on U.S.-China Relations.
