Reviews surfacing on social media ranged from sensational to lacklustre.
“Got access and it’s true... Manus is the most impressive AI tool I’ve ever tried,” Hugging Face’s head of product design Victor Mustar said in a post on X.
“The agentic capabilities are mind-blowing, redefining what’s possible.”
Criticism included those saying Manus stumbles on simple tasks such as booking a flight, or that they ran into error messages or endless loops.
And since the AI processing is hosted in the cloud, users worried about the security of their data.
Whether Chinese companies are taking the lead on AI has been a hot topic since China-based DeepSeek burst on to the scene in January.
DeepSeek’s model challenges those created by OpenAI, Google, and other US rivals but operates at a fraction of the cost.
The latest artificial intelligence trend has been digital “agents” specialised for specific tasks or fields.
Anthropic and OpenAI have both added such capabilities to their AI platforms since late last year.
Butterfly Effect described Manus as being able to carry out tasks such as buying property in New York or editing a podcast.
But TechCrunch journalist Kyle Wiggers wrote of Manus failing when asked to order him a sandwich or find him a plane ticket to Japan during a try-out.
China’s rapid advances in AI despite US restrictions on exports of cutting-edge computer chips worry Silicon Valley.
And unleashing AI agents on the internet without tight regulation raises concerns about mishaps or abuses, like stock market chaos caused by digital agents that make factual errors.
Corpora.ai chief executive Mel Morris did not see Manus as a “revolutionary leap” from existing AI models but saw its ability to access remote computer servers as a potential risk to data confidentiality.
-Agence France-Presse
