A brand new artificial intelligence named Manus is causing a little bit of a buzz online. Invitation codes are selling for steep prices. Some draw comparisons to ChatGPT’s early rush, believing it might mark a new milestone for agent-based AI. Some voices call it a major jump in automated workflows.
Manus was unveiled by The Butterfly Effect developers in China. Invitations are scarce, which has only made it more alluring. Rumours swirl about its ability to handle everything from real estate searches to personal errands with minimal oversight.
It also reminds folks of DeepSeek, another Chinese AI that turned heads in the past. Others are guarded, since it’s still under development and prone to errors.
How Is Manus Different From ChatGPT And Similar Tools?
Manus works in the background and gives updates only when tasks finish. Users see it tackling real jobs, like sorting schedules or scanning property ads, without constant input. Its ability to take on multi-stage projects distinguishes it from simpler chatbots, freeing people from repeated prompts.
Its developer, Yizhao “Pika” Ji, points to strong performance on the GAIA benchmark. That has stirred curiosity, since it hints at more than basic Q&A functions. Ji believes this signals a new direction for AI usage, since it tackles daily needs rather than mere conversation.
Manus acts more like an operator than a talk-based tool, bridging thought and action behind the scenes. The team calls it a complete agent that gets results, rather than mere conversation.
[skynws]
What Practical Examples Are People Seeing?
Fans praise its skill in trip planning. One demonstration showed a detailed holiday outline for Japan, plus a personalised guide brimming with local tips.
Another example involved stock analysis for Tesla, where Manus produced charts of share swings and possible scenarios. Testers say this cuts the need for repeated instructions.
Some also describe how Manus combs through big databases to arrange results into structured tables. People find that more flexible than standard chatbots, which usually demand line-by-line commands.
In another test, it prepared slides and scripts on the momentum theorem for teachers, demonstrating its knack for educational tasks. Sceptics warn that it might overlook details, so real-world reliability is still unproven.
What Still Needs Work?
Early testers speak of hiccups that disrupt tasks. A startup co-founder reported that the system fell into endless restarts. Others mention factual blunders, or claims that skip important data. One user on X also saw it omitting entire chunks of research when asked for references.
Some on X say it misquotes information or misses obvious points. These stories discourage certain users from trusting it with important assignments.
Developers confirm they are gathering feedback during a closed beta. They promise to fix bugs and refine each feature. Opinions are split on when it might be robust enough for large-scale use.
Where Does Manus Go From Here?
The Butterfly Effect team plans to add more computing resources for the influx of testers. Their Discord server has already passed 138,000 members, and invitation codes fetch steep prices on Chinese platforms.
High-profile figures have shared strong reactions. The Hugging Face product lead called Manus “the most striking AI tool” he has tried, and an AI policy researcher called it “the most complex computer using AI.”
Manus could go a similar route, or it might prove that an autonomous agent can thrive right now. Many marketers see business value if it stabilises, but no one knows if it can meet the grand talk that has built up around it. Brand managers see a chance to start using it for chat support or scheduling interviews, hoping to stay ahead of rivals.
