Athira Sethu
Kochi, 10 March 2025
A fresh AI model out of China, Manus, is creating a buzz in the tech community. Only weeks after the DeepSeek AI model made headlines, Manus is being pitted against the best AI models out of OpenAI, Google, and Anthropic. It’s receiving accolades for being a robust agentic model, which can perform various tasks on its own.
Manus was created by Monica, a Chinese startup company. The developers call it an AI that “bridges minds and actions,” i.e., it not only thinks but also produces results. Manus can independently carry out a variety of real-world tasks. From creating websites to organizing trips and analyzing stocks, this AI system can do everything with a single command.
Launched on March 6, Manus has already caught global attention, with many praising its ability to outperform OpenAI’s DeepResearch in specific benchmarks. Unlike regular AI chatbots, Manus is a general-purpose AI agent capable of planning, executing, and completing complex tasks. For example, if asked to write a report on global warming, Manus would handle everything, from researching the topic and writing the paper to generating charts and interactive elements, completely on its own.
Manus is also distinguished by its live interaction and workflow visualization. During a demo, the AI demonstrates how it interacts with the web, collects data, and carries out tasks in real-time. For example, when asked to plan a day-by-day Japan itinerary, Manus not only developed the itinerary but also visualized its process in real-time.
One of the most impressive aspects of Manus is that it can function independently. Once a user assigns it a task, it is able to keep working in the cloud even if the user logs off. It can access platforms like X and Telegram, capture screenshots, monitor browsing history, and create files, such as PDFs or spreadsheets.
Another aspect is its ability to personalize. Manus learns from user interactions and can customize its outputs according to personal preferences.
Manus is currently available as an invite-only preview. The firm has not yet stated when it will be released publicly, but it will open-source the model later this year so developers can add it to their own work. Users can currently engage with Manus in the same way they would other AI systems such as ChatGPT, by beginning with a prompt and allowing Manus to do the rest.
