Dim Amor
In February 2026, Meta’s artificial intelligence system began appearing on the Facebook platform under the name "Manus AI", a move that illustrates the integration of technology acquired at the end of 2025 and the company’s intention to lead the field of autonomous agents. This marks another step in the artificial intelligence race among tech giants, amid competition with companies such as Google and OpenAI.
At the end of 2025, Meta announced the acquisition of the AI startup Manus in a multibillion-dollar deal. According to a report in The Wall Street Journal, the transaction exceeded two billion dollars, although the company declined to disclose the full terms. The goal of the acquisition was to integrate Manus technology into Meta’s AI products, particularly the Meta AI systems operating across Facebook, Instagram, and other services.
Manus is an artificial intelligence platform developed in 2025 by the company Butterfly Effect. It represents a new generation of systems known as "autonomous agents". Unlike traditional chatbots that simply answer questions, an autonomous agent is designed to receive a task and carry it out from start to finish independently.
The core idea behind the system is a shift from AI that merely provides information and explanations to AI that performs real actions. Manus was designed to execute complex real-world tasks without continuous user guidance. Possible tasks include research, code writing, data analysis, website creation, and travel planning.
The difference between an autonomous agent and a conventional chatbot lies in how they operate. Conversational systems follow a basic pattern: the user asks a question, the system responds, and then waits for the next prompt. In contrast, an autonomous agent receives a goal or task, builds an action plan, uses digital tools such as a browser, code, or documents, performs the task independently, and ultimately returns a completed result.
For example, instead of asking for instructions on how to write a report on the automotive market, a user could assign the system a task: prepare a full report on the European automotive market. The system would then carry out the research, data collection, analysis, and writing stages automatically, delivering a finished product.
The system was designed to operate in the cloud even when the user is offline, enabling it to perform complex tasks without continuous intervention. Key capabilities include planning and executing complex assignments, working with text, images, and code, conducting research and data collection, analyzing information, and producing reports.
Manus operates from a Singapore-based platform with roots in China. The company offered paid subscriptions to customers interested in using the technology for research, programming, and other tasks. According to company statements, the platform surpassed 100 million dollars in annual recurring revenue just eight months after launch.
Early investors included Tencent Holdings, ZhenFund, and HSG. The developer company, Butterfly Effect, was founded in China and later moved its main operations to Singapore. Following the acquisition, Meta stated that Manus would no longer have any Chinese ownership interests and that the platform would cease operations in China. The main operational center is expected to remain in Singapore.
In a statement, Meta said Manus already serves millions of users and businesses worldwide, and that the company intends to expand the service to provide general-purpose agents across all of its consumer and enterprise products, including Meta AI.
Manus CEO Xiao Hong said that joining Meta would allow the company to build on a stronger foundation without changing the platform’s operating model or decision-making processes.
Alongside its advantages, the autonomous agent model also raises professional, ethical, and legal questions. One of the main challenges concerns the degree of user control over the process. While conversational systems allow adjustments and guidance at every stage, an autonomous agent operates independently and may sometimes perform actions that were not precisely defined.
Another challenge relates to information reliability. When the system selects sources, interprets data, and produces a final output without human oversight at every step, there is a risk of factual errors, reliance on unreliable sources, or incorrect conclusions. In addition, the autonomous agent’s operation includes web searches, code execution, and file creation, raising issues of transparency and oversight.
In sensitive fields such as journalism, law, and investigative reporting, a system that operates autonomously may collect inaccurate information or formulate legally problematic claims due to the absence of human review at each stage. By contrast, conversational systems allow full control, precise wording, and verification before publication.
Nevertheless, autonomous agents offer a key advantage: the ability to perform complex tasks independently and save time in technical or business projects. Potential uses include website development, advanced data analysis, process automation, and the production of comprehensive reports.
The acquisition of Manus is part of a broader effort by Meta to strengthen its artificial intelligence operations. During 2025, the company invested 14.3 billion dollars in the data company Scale and recruited its CEO, Alexandr Wang, to help lead the "superintelligence" team.
The appearance of "Manus AI" on Facebook in February 2026 marks the beginning of the integration of autonomous agents into large-scale consumer platforms. For users, this could mean a shift from systems that only provide answers to systems capable of carrying out entire tasks independently.
The acquisition and integration into Meta’s platforms point to a broader shift in the artificial intelligence industry: a transition from conversational chatbots to autonomous agents designed to perform real-world actions. It represents another stage in the technological race among major AI companies, each seeking to lead the next generation of intelligent systems.
