Weeks after Chinese AI startup DeepSeek caused a whirlwind with its DeepSeek-R1 model, China seems to be bracing for another milestone moment in the AI arms race. Another startup, Monica, has released its artificial intelligence (AI) agent called Manus. Just as seen with DeepSeek, Manus is currently attracting widespread attention worldwide. Some publications have already monikered it as ‘yet another DeepSeek moment’ for China.
Manus is reportedly grabbing eyeballs owing to its ability to handle complex real-world tasks. The AI agent, which is currently available as an invitation-only web preview, is capable of executing practical tasks such as creating websites through a step-by-step process. The company has shared a demo video of the AI agent on its official website—manus.im. Reportedly, the AI agent went viral in less than 20 hours since its preview was launched.
What is Manus?
Manus is being dubbed as the world’s first truly general AI agent. Reportedly, the AI agent has demonstrated its capabilities, such as thinking, planning, and executing tasks independently, delivering complete results, etc. According to the website, the Manus AI agent also has the ability to come up with an itinerary for a trip to Japan, offering an in-depth analysis of stocks, creating interactive coursework for school teachers, analysing different insurance policies, etc. According to the company, Manus can also outperform OpenAI’s DeepResearch based on the GAIA benchmark—a third-party benchmark to assess general AI assistants.
Chinese startup Monica just launched Manus, the world’s first general AI agent. #Manus has set new state-of-the-art (SOTA) performance across all three difficulty levels, surpassing #OpenAI’s Deep Research in the GAIA benchmark for real-world problem-solving capabilities. pic.twitter.com/eiRShk1hxq
— Chengdu Hi-tech Zone (@ChengduHitech) March 7, 2025
According to the makers, Manus AI has been designed to autonomously perform complex tasks across various industries. Its distinct features are autonomous operations and asynchronous cloud-based functionality, which allows users to assign it tasks and disconnect their devices. The makers claim that Manus, even after being disconnected, continues working in the cloud and provides results upon completion. The AI agent adapts to the user’s preferences over time, enhancing its performance and personalisation and continued use.
Story continues below this ad
The company shared the demo video on X on Wednesday night, and it gained over 200,000 views by noon Thursday. Several users were seen requesting invitation codes to access the AI agent. Weeks ago, a similar sense of excitement was seen around DeepSeek R1, DeepSeek AI’s model that was reportedly made at a fraction of the cost of the frontier models by OpenAI, Meta, Google, etc.
Based on the demo, Manus was seen interacting with its environment, gathering data, executing tasks, browsing websites, utilising various tools, and displaying its workflow in real time. Ever since Manus made it to the internet, several users have been sharing clips of its jaw-dropping demonstrations. One of the demos shows the AI agent controlling 50 different screens simultaneously. The AI agent can be seen using various online platforms like X, Telegram, and many others all at once in a demonstration of unparalleled level of automation.
今天被Manus刷屏。通过制定不同的agent（工作流）不断的被llm调用，然后执行各种各样的工作任务。其实本质上依然是一个人缝合怪，但是的确做的非常漂亮。
对于大部分普通人来说，在庞大的agent工作室里面找到一个适合自己来进行工作的难度是不小的。Manus很好的解决了这个问题。… pic.twitter.com/iiajq3Z030
— Leon.Y (@StlightLeon) March 6, 2025
So far there are not many details about the company. However, the demo video was hosted by Peak Ji Yichao, a 33-year-old Chinese entrepreneur and tech enthusiast who is known to have created the mobile browser named Mammoth.
How to use Manus AI?
Getting started with this new AI is fairly simple. Like in another AI chatbot, one can begin by using a prompt. For example, Give a 7-day travel itinerary with a specific budget to Bali. Following the prompt, Manus researches and compiles data, creating structured output in real-time. Manu AI can generate files and actively browse the web for information. It can have live interaction with web pages, take screenshots, and even record browsing activity for later review.
A new dawn for AI-powered browsing?
Tech companies have been beelining to launch their AI-powered browsers, signalling a broader industry shift towards intelligent web navigation. Manus AI seems to be a new dimension in this movement as it offers a more intuitive, powerful, and cost-effective alternative to existing tools.
Story continues below this ad
With its ability to conduct deep web research, create files, and automate an array of online tasks, Manus AI may likely change how we interact with the internet. This rapid pace of Chinese AI development suggests that Manus could be the next big thing in AI-driven automation.
