Hype or Breakthrough? Chinese startup unveils Manus, the "world's first" fully autonomous AI agent

A team of Chinese developers from The Butterfly Effect has unveiled the "world's first" fully autonomous artificial intelligence agent called Manus. Unlike traditional AI chatbots like ChatGPT, Google's Gemini, or xAI's Grok, which require human input, Manus is capable of making decisions and performing tasks on its own without constant supervision, TechCrunch reports.

Manus launched in early access last week, and is invite-only, but it has already caused a stir comparable to the launch of another Chinese AI from DeepSeek:

• Hugging Face's head of product called Manus "the most impressive AI tool I've ever tried";
• AI policy researcher Dean Ball stated that it is "the most complex computer using AI";
• The official Manus Discord server has surpassed 138,000 members in a few days, and invitations to the platform are selling for thousands of dollars on the Chinese marketplace Xianyu.

Unlike traditional AIs that work on a request-response basis, Manus performs complex tasks in the background and notifies the user only when the work is complete.

For example, if a user asks Manus to find an apartment, the AI can analyze the real estate market, crime rates, and climate conditions, as well as assess financial feasibility and provide personalized recommendations.

According to Manus developer Yizhao "Pika" Ji, Manus outperforms OpenAI's Deep Research and Operator in the GAIA benchmark, which tests AI's ability to work with browsers, software, and perform complex tasks.

"Manus is not just another chatbot," Ji said. "It’s a completely autonomous agent that bridges the gap between conception and execution [...] We see it as the next paradigm of human-machine collaboration."

But is Manus really that advanced? Despite the considerable hype, early testers are reporting serious flaws. Oleksandr Doria, co-founder of the startup Pleias, said that during testing Manus encountered errors and endless reboot cycles. At the same time, many users on X (Twitter) write that Manus makes factual errors, does not always correctly cite sources and often omits obvious information.

A Manus representative said in a comment to TechCrunch:

"As a small team, our focus is to keep improving Manus and make AI agents that actually help users solve problems. The primary goal of the current closed beta is to stress-test various parts of the system and identify issues. We deeply appreciate the valuable insights shared by everyone."

Despite the criticism, Manus developers say they are working to scale up computing power and fix issues. However, at this early stage of development, Manus still looks more like an experiment than a finished technological product.

Manus may be a game-changer in artificial intelligence in the future, but for now it's more of a proof-of-concept than a fully functioning AI agent.