(Yicai) March 12 -- Manus, an artificial intelligence agent developed by Chinese startup Butterfly Effect, has linked arms with the team behind Alibaba Group Holding's Qwen artificial intelligence models to build the agent's domestic version.

Manus and Alibaba's team aim to fully integrate the agent's functions into Chinese AI models and computing platforms, the former announced on its website yesterday. Tech teams from both sides have already started close cooperation, it added.

Manus sparked widespread interest in the AI sector after the release of its early preview on March 5 after Wuhan-based Butterfly Effect claimed it was the world's first fully autonomous AI agent. According to a demo video, users can give simple commands to complete complex tasks, including resume screening, property selection, and stock analysis.

After the deal with Alibaba, Manus' website redirects users to a page reading, "The Chinese version of Manus is under development, powered by Tongyi Qianwen's dedicated model." On March 8, the site switched to a Chinese interface that supports multiple languages, including English, German, Italian, French, and Portuguese, from an English-only version.

"Manus offers the public a glimpse into an intelligent assistant capable of integrating multiple tools, showcasing a novel product concept," The Paper reported yesterday, citing Tian Feng, dean of the intelligent industry research institute of loss-making Chinese AI startup SenseTime Group.

As foundational LLMs adopt modular integration, application programming interfaces and open-source tools become cheaper and more accessible, accelerating product innovation, which could lead to a surge in consumer-facing toolkits, Tian said. Unlike DeepSeek's underlying technology, Manus is easier to implement, he added.

OpenAI also launched new agent tools late yesterday to streamline relevant development, enabling developers and companies to build reliable and practical AI agents effortlessly. The new releases include the open-source Agent SDK, designed to simplify the creation and management of complex single-agent and multi-agent systems, and the Responses API, optimized for agents while retaining all functionalities of the firm's existing Chat Completions API.

Editor: Martin Kadiev