Click “Shanghai Financial Information Industry Association” to follow us
On March 12, OpenAI released the Agent Toolkit, aimed at simplifying the development process of agents, improving development efficiency, and building a platform for agent developers. Just a day earlier, the recently trending Manus announced a strategic partnership with Alibaba’s Tongyi Qianwen team, dedicated to creating more creative general-purpose intelligent agent products for Chinese users. There have been opinions suggesting that 2025 will be the year of the commercial explosion of AI Agents. Now, this prediction seems to be coming true.
Manus’s Reputation is Polarized, Market Enthusiasm is Ignited
Claimed to be the “world’s first general-purpose agent,” Manus is a multi-model, multi-agent product. Unlike traditional agents that are designed for single tasks with strict rules and processes, Manus possesses the ability to “think independently,” capable of breaking down and executing complex general tasks in multiple steps.
Some believe that the reason Manus has shocked many is that it demonstrates a more advanced self-thinking and self-planning capability. Users do not need to guide the conversation or set specific scenarios; a single sentence can initiate complex tasks across different fields, upgrading AI from a “suggestor” to an “executor,” achieving a paradigm shift from “passive response” to “active execution.” This is a victory of systematic engineering.
However, some opinions suggest that compared to DeepSeek, Manus’s technological value is not high, with more marketing than innovation. “This is like the first homework after the first class in the agent development world. The key to AI Agents is the know-how knowledge base. Many people are working on general AI Agents, and everyone has the know-how; there is no gap based on large language models. But creating an AI Agent that can build rockets is difficult because there are few people who understand rocket building.” An industry insider commented. Others have stated that after this wave, “changing the world has not yet happened, and the magic has not yet occurred.”
In response to various external doubts, Manus product manager Zhang Tao stated: “Manus is still a baby in swaddling clothes, far from the experience we want to deliver in the official version. There is still much room for improvement in areas such as model hallucination, deliverable friendliness, and operational speed.”
Multiple Players Entering the Market, Competing for Commercialization Opportunities
Despite mixed reviews of Manus in the market, one clear point is that the industry has reached a consensus that 2025 is likely to be the year of the explosion of AI Agent applications. Market research firm Gartner has listed Agentic AI as one of the top ten technology trends for 2025, predicting that by 2028, at least 15% of daily work decisions will be made autonomously by Agentic AI. Deloitte’s AI Research Institute states that AI agents will change fundamental business models, realizing new ways of work, operations, and value delivery.
Whether it is large model startups, major internet companies, software application firms, or operators, they are all choosing to join this race, hoping to benefit from new traffic entry dividends. In the international market, OpenAI has launched its first AI agent “Operator” and an agent for deep thinking “Deep Research” since the beginning of this year, planning to further validate the B-end potential of AI Agents. Microsoft has launched ten AI Agents at once and announced the establishment of the world’s largest enterprise-level AI Agent ecosystem – the Copilot Studio platform. Google quickly followed suit by launching a commercial AI Agent market, providing a one-stop commercial ecosystem from development to deployment and application of AI Agents. Amazon has reportedly formed a new team dedicated to Agent development.
Looking at the domestic market, most major internet technology companies have chosen to provide platform-based, one-stop intelligent agent services, competing for the opportunity for large-scale implementation. For example, Baidu’s Wenxin intelligent agent platform has now gathered 150,000 enterprises and 800,000 developers. Alibaba’s strategic cooperation with Manus also reveals both parties’ ambition to accelerate the industrialization of general intelligent agents. ByteDance positions Coze (Button) as an intelligent agent creation platform and has released it across its platforms such as Doubao and Feishu, rapidly expanding its user base.
In contrast to major internet companies, AI startups tend to combine their own model capabilities with intelligent agents to launch more scenario-based implementation solutions, seeking paths for commercial monetization. For instance, Zhipu AI, one of the earliest startups exploring agents, launched its autonomous intelligent agent AutoGLM last October, including three versions: AutoGLM, AutoGLM-Web, and GLM-PC, corresponding to mobile, browser, and computer application scenarios. Jietian Xingchen announced its focus on intelligent terminal agents, deepening partnerships with Geely Automobile Group, OPPO, and Zhiyuan Robotics, and focusing on innovation and application in finance, content creation, new retail, and digital humans.
Embrace Diversity, Pursue Excellence
Open-minded and Wise, Magnanimous and Humble
Shanghai Financial Information Industry Association
Official website: http://sfia.org.cn/
We welcome everyone to scan the QR code and follow us