China seems to be on a roll with its new AI models as it has just launched Manus as the new AI agent just after DeepSeek took the world by a storm. Monica – a Chinese startup – has launched Manus as an agent that can bridge the gap between the mind and action. Dubbed as an AI agent that can deliver results, rather than simply think, people are losing their minds on how Manus can perform some complex tasks that AI agents could not do in entirety without assistance. Such is the sophistication of the model that it performs better compared to OpenAI’s DeepResearch model on the GAIA benchmark.
Why is it creating buzz?
A video that has gone viral shows Manus performing several tasks in parallel, which can be visualized on a desktop screen. With capabilities to book travel itineraries for people, to analysing stocks and creating coursework for educators, Manus is competent enough to take care of tasks end-to-end with just a single prompt. What sets it apart from existing AI models is that instead of simply providing answers to users’ queries, Manus can take full responsibility of completing tasks for humans that would otherwise consume time and human labour. For example, if you want other AI models to prepare a wedding invitation card for you, the models may prepare the text, or suggest some ideas for designs. Manus takes this several notches higher by composing the invitation, generating the design, compiling all information pieces into a full-fledged document and making it ready for dispatch, with just a single prompt. It can even create custom websites for you, even if you carry zero experience in programming.
What’s the story behind Manus?
This general AI agent has been built by a low-profile Chinese team which is backed by Chinese investors and is allowing people to use the model on invite-only basis. It operates independently in the cloud and therefore, can continue working even if the user disconnects its device, thereby enabling Manus to go uninterrupted in its long-term assignments. What makes Manus even better is its capability to reflect on its responses, adapt to user preferences, and improve the quality and relevance of its output. Since the model can work in unsupervised fashion, many users are dubbing it as the first ever general AI. In one of the viral videos, Manus can be seen using almost 50 screens and utilizing them to operate platforms, such as X, Telegram and others to fetch information and execute commands. By taking screenshots, generating files and recording browsing history, Manus has shown it can do a lot without guidance from humans. Manus also showed it could create day-by-day itinerary for a trip to Japan with just a single prompt from the user. Moreover, it can work seamlessly in the background, while the user attends to other more important tasks which require higher cognitive skills.
Not everyone is excited though
Diverse opinions always exist, and Manus’s case is no different. The co-founder of Pleias – Alexander Dora – tweeted on X that he could spot Manus encountering errors and entering infinite loops during testing phase. Some users also highlighted that Manus could be found faltering on a few factual questions and the citations were not consistent. Some of the information and knowledge that could be easily found online were not fetched by Manus in certain cases. At times, when users asked Manus to find and book flights and other services, it returned outdated and broken links, while failing to make reservations at restaurants in a separate case.
Though China has touted Manus as the next big thing in the AI space, the creators of Manus have clarified that they intend the current closed beta version to stress-test various parts of the system and highlight existing issues, and therefore, improvements are underway. The company has also clarified that it is working to scale the compute capacity and fix the gaps as they are reported by users.
However, if issues persist and more issues surface, industries won’t be able to blindly rely on Manus for mission-critical action items. No wonder, the moment any new AI model steps into the industry, the ecosystem of influencers goes crazy and hype the model to be the next gamechanger. With Manus entering the scene, it is yet to be seen whether it stays as a trendsetter in the general AI space or is soon replaced by another sophisticated model in a matter of weeks.
