A New AI Entity Is Born: Manus AI Sparks Attention and Skepticism
BEIJING — A new name in artificial intelligence, Manus AI, made its mark on March 5, 2025, creating a buzz north and south of the Great Wall and beyond.
Created by the startup Butterfly Effect, this “general AI agent” claims to be able to perform tasks independently, like organizing resumes or booking trips. It launched with a viral video, and is generating buzz — but not everyone is sold on the hype.
A Sweeping Assessment Draws a Mixed Response
At Manus AI, we hit the ground running. Its creators — a small firm with offices in Beijing and Wuhan called Butterfly Effect — refer to it as the world’s first fully autonomous AI agent. Whereas chatbots require continuous prompting to stay on topic, Manus can start and complete complex tasks with a single prompt. A demonstration showed it ranking job applicants and creating a website in minutes. On March 13 it had over 138,000 members on its Discord server.
The enthusiasm is not universal. Early testers encountered hiccups — crashes and loops and wobbly outcomes. But others say it’s riding the coattails of models already out there, like Anthropic’s Claude and Alibaba’s Qwen, not novel tech. Skeptics ask if it’s more flash than substance. Still, its fans are comparing it to DeepSeek, a Chinese A.I. that stunned the world in January.
Widely hyped, but what’s actually going on with Manus?
This story is powered by China’s AI race. The U.S. dominates in big names like OpenAI, but China is rapidly catching up. The low-cost model used by a new AI friend DeepSeek sent investors reeling, costing Nvidia $600 billion in stock. With that, Manus comes in, making the claim to out-perform OpenAI’s Deep Research on the GAIA benchmark. That’s a test of real work,” Manus boasts it’s better.
The Tech Breakdown
How does it work? Reccia combined several AI models to make him a one-man show. It runs in the cloud, so users can log off while it chugs away. “It’s not just a chatbot — it’s a doer,” Yichao “Peak” Ji, Butterfly Effect’s co-founder, said in the launch video. It’s playground for tasks such as stock analysis or hunting for an apartment. But constraints exhibited: it fumbled on simple commands like scheduling flights.
The Skeptic’s View
Not everyone’s cheering. Kyle Wiggers of TechCrunch gave it a try, and he found faults. It couldn’t place an order for a sandwich or make a flight reservation. Privacy concerns also bubble up. Butterfly Effect is legally based in Singapore, but the team is in China. Where’s the data going? Experts like the journalist Luiza Jarovsky warn of risks if the servers connect to China.
A Tricky Landscape and Grand Promises
Manus isn’t alone. Such “agentic” AI — tools that act rather than just jabber — is available from OpenAI, Google and others. OpenAI’s Operator and Anthropic’s Computer Use mode move users sliders and clicks. Manus is claiming more: complete autonomy. Butterfly Effect calls it a glimpse of artificial general intelligence, or AGI — A.I. that thinks like humans do.
That’s a bold leap. DeepSeek constructed its own models and made them available. Manus conceals its recipe exercising suspicion The invite-only accessibility elicits a sense of intrigue — and irritation. On China’s Xianyu app, invite codes sell for thousands of dollars. Zhang Tao, a product partner, apologized on social media, blaming server woes. “It’s still a baby,” he said.
What’s Next for Manus AI?
The road ahead is foggy. Butterfly Effect pledges to eliminate bugs and scale. The faster Manus delivers, the more industries it could upend — from quicker hiring to more astute investing. But if it bombs, it’s another overhyped story. China’s AI effort won’t halt regardless. China may turn up the pressure on U.S. as Beijing flexes tech strength.
This debut matters. It is a test of trust and tech. So those are the motivators behind Manus, but will it come into its own, or will it turn out to be just a flashy demo? We will see over the next few months. For now, it’s an issue — and an issue a hot one.