- The new original AI tool manusion is praised as more capable than agents previously available
- Others are less impressed, citing accidents and hallucinations, while experts call for caution
- A user gallery shows that manus is used to plan trips and analyze actions
Manus, a new “ agentique ” ‘tool from the Chinese startup Monica, concluded an overview only of the invitation and, according to some users, “[redefines] Which is possible ”for automated AI automated workflows to be carried out.
It is according to a Twitter user who asked Manus to “code a Threejs game where you control an airplane” and “ended up concluding that manus is the most impressive AI tool I have ever tried”.
The own manus use case gallery (via Techcrunch) has some of the possibilities, such as traveling a trip and the production of a personalized guide for a trip to Japan, access to a database to sort computer information in tables, analyzing the stocks of a real person.
What are AI agents?
The “ agents ”, although propelled by models of great language (LLMS), differ from writers of AI in that they use systems to automate a workflow according to the prompt of a user, rather than producing a single output element.
Manus does not offer anything particularly new, although it is remarkable that, so far, perhaps, AI agents have not had the best reputation; Known for having fought with factual and essentially non -functional information.
Others, however, are not as impressed. TechcrunchKyle Wiggers gave a coherent report of their experience, signaling accidents and failures on tasks ranging from simple (reserving a table in a restaurant) to the alcohols (“then I asked the platform to build a fighting game inspired by Naruto)”.
Comparisons have also been made with the LLM of Deepseek, although it does not add quite. Manus and Deepseek could both from China, but Deepseek is not an agent, and manus is not open source and has no release calendar. That said, an open source version, Openmanus, is in active development.
Manus from a Chinese startup caused discomfort, with circles, Axios Notes, China is worried about the ground in a so-called “race for AI supremacy”.
Experts Doubt Manus
Two industry personalities told Techradar Pro that their doubts surrounding the AI agency still remained. CEO of Corpora.ai, Mel Morris, said that “manusal manifestations” present the potential of agentic AI as a significant step forward. However, granting it from autonomous AI agents as a manus, the ability to carry out independent actions raises serious concerns. »»
“We have already attended many examples of generative AI and so -called reasoning models producing erroneous outings. If we give autonomy on high stake tasks – such as the purchase and sale of actions – such imperfections could lead to chaos. The ability to compare the fundamentals of actions and negotiation activities in minutes is very different from successful navigation in market complexities. »»
Likewise, Copyleaks’ co-founder and CEO, Alon Yamin, said: “Although IA innovation can stimulate remarkable efficiency, it also increases concerns about transparency, data security and responsible monitoring. Manus -based cloud operations surface several critical questions about data governance: who controls information? Where does he reside? How is the data protected? »»
“As AI systems become more autonomous, ethical development and regulatory alignment must be prioritized. AI governance solutions that help organizations to detect and mitigate the risks of AI remain essential, guaranteeing progress like manus is responsible for responsible and transparent. »»
