‘Less structure, more intelligence’: AI agent Manus draws upbeat reviews of nascent system


General-purpose artificial intelligence (AI) agent Manus has drawn a wave of mostly upbeat reviews, days after Chinese start-up Butterfly Effect made it available through an invitation-only online preview that triggered comparisons of a DeepSeek-level tech breakthrough.

“Got access and it’s true ... Manus is the most impressive AI tool I’ve ever tried,” wrote Victor Mustar, head of product at AI and machine-learning developer platform Hugging Face, in a post over the weekend on X, formerly Twitter.

Mustar prompted Manus to create and display an animated 3D video game in a web browser, using a cross-browser JavaScript library and application programming interface – known as Three.js.

Do you have questions about the biggest topics and trends from around the world? Get the answers with SCMP Knowledge, our new platform of curated content with explainers, FAQs, analyses and infographics brought to you by our award-winning team.

His prompt: “code a threejs game where you control a plane”.

That simple prompt shows how AI agents like Manus represent a new avenue for innovation, compared with chatbots and large language models (LLMs) – the technology underpinning generative AI assistants like ChatGPT.

AI agents are programs that are capable of autonomously performing tasks on behalf of a user or another system. Essentially, these agents create a plan of specific tasks and subtasks to complete a goal using its available resources.

“The UX [user experience] is what so many others promised ... but this time, it just works,” Mustar said in his post.

A demonstration video online at manus.im shows the AI agent executing various practical tasks, such as creating a custom website. Photo: Shutterstock

Some of the initial reviews of Manus in mainland China have also been positive.

On Quora-like question-and-answer service Zhihu, Li Guohao – founder of multi-agent framework and open-source community Camel-AI.org – lauded Manus’ system design for achieving “less structure, more intelligence”.

Following Manus’ preview, Li expected another “aha moment” in AI agents within a year.

Chinese tech blogger Lan Xi, who has nearly 3 million followers on microblogging platform Weibo, used Manus for five tasks – including creating online games and generating repair tips for a Boston Dynamics robotic arm. Manus failed one task because it could not sign up for an online drive to obtain images for a video game, according to Lan.

The blogger said Manus has the potential of becoming a “DeepSeek V2 moment” in the AI agent segment of the industry.

In a video posted on his WeChat channel, Chinese investor and entrepreneur Fu Sheng dismissed recent comparisons between Butterfly Effect’s Manus and AI model developer DeepSeek because they are involved in different areas of the AI sector.

“DeepSeek represents the core of intelligence, while Manus is more like an additional layer on top of a large language model,” he said.

That makes Manus significant because it makes LLM “more accessible”, according to Fu, who also serves as chief executive of New York-listed internet firm Cheetah Mobile. “It’s like how Microsoft’s Windows operating system became widely adopted because of its user-friendly interface,” he added.

On Chinese social-media Jike, Wang Junyu – co-founder of Chinese app publisher Wandoujia – said Manus was “clearly better” than other existing AI agents in the market such as OpenAI’s Deep Research. He indicated, however, that Manus appeared to be similar to Devin AI from US start-up Cognition Labs.

Chinese tech media Synced reported that two out of five tasks it ran on Manus failed. One was because of a sign-up issue, while the other involved a task to streamline the firm’s page layout, which did not work properly.

In a post last week on Jike, Manus product partner Zhang Tao wrote that “the current version of Manus is still in its infancy, far from what we aim to deliver in our final product”.

Butterfly Effect co-founder and chief scientist Peak Ji Yichao said Manus was built on existing LLMs, including Anthropic’s Claude and fine-tuned versions of Alibaba Group Holding’s Qwen AI models, according to a Monday post on X. Alibaba owns the Post.

More from South China Morning Post:

For the latest news from the South China Morning Post download our mobile app. Copyright 2025.

Get 20% OFF The Star Digital Access

Monthly Plan

RM 13.90/month

RM 11.12/month

Billed as RM 11.12 for the 1st month, RM 13.90 thereafter.

Best Value

Annual Plan

RM 12.33/month

RM 9.87/month

Billed as RM 118.40 for the 1st year, RM 148 thereafter.

Follow us on our official WhatsApp channel for breaking news alerts and key updates!

Next In Aseanplus News

Crash of Indonesian water-bombing helicopter kills crew member
Philippines rebukes officials over public comments about China
Ahmad Zahid's working visit to Turkiye strengthens diplomacy, explores new opportunities
Australian shares slip below 9,000 level as banks overshadow miners' gains
Pre-installed XCPE malware targets Android TV boxes, Nacsa warns
Oil falls on open as Trump says fresh Iran attacks 'unlikely'
RBA's Hunter says AI has yet to deliver productivity boost, trigger job losses
Nvidia opens probe after report that CEO Huang was target of foreign intelligence
AI slump revives Asia’s overlooked markets
Lawyers: No reason to keep Najib in Kajang Prison

Air Pollutant Index

Note: API is calculated based on 24hr running average of the most dominant parameter.

Highest API Readings

    Select State and Location to view the latest API reading

    Others Also Read