The next wave of AI transformation will be driven by agents. This next-generation AI surpasses chatbots like ChatGPT by not simply generating responses but by taking action.
Unlike chatbots, these systems carry out complex tasks with little human involvement. In coming years, we may see them evolving into virtual assistants, teachers, employees, carers and even friends.
Right now, everyone’s waiting for AI agents to have their “ChatGPT moment”—when millions of people across the world suddenly realize it could actually be very useful. Now, in the wake of the market turmoil caused by DeepSeek, a new Chinese contender has emerged.
The developer of Manus AI claims it’s “the first AI agent … a truly autonomous agent that bridges the gap between conception and execution” and “potentially a glimpse into AGI”.
So, what is it, what does it do, what are its political implications in a world where superpowers are racing for AI supremacy? And is it really a step towards the AI holy grail of AGI?
So, What Is Manus?
Manus is currently an invite-only research preview, immediately casting doubt on developer Monica’s claim that it’s the first AI agent. OpenAI’s similar Operator agent model is also a research preview but is at least available to customers – if only for those paying for the $200-per-month ChatGPT Pro tier.
But a developer video has been released showing it performing real-world tasks like screening resumes, researching housing and analyzing stocks.
Its name, derived from the Latin “Mens et Manus” (“mind and hand”), reflects its structure: LLM algorithms as the mind and deterministic algorithms as the hand. Unlike ChatGPT, which only generates responses, Manus can execute actions—integrating with services, processing data, and performing operations like a traditional computer program.
For example, ChatGPT can write Python code, but Manus can autonomously execute, test, and refine it. Instead of summarizing financial reports, it fetches real data and runs its own analysis.
Interestingly, its creators say that coding doesn’t have to be a goal for Manus, which actually thinks of it as a “universal tool for solving problems.” This suggests Manus could autonomously write and execute code to address challenges we’ve never thought to tackle through programming—expanding AI’s problem-solving horizon beyond our current imagination.
One other important point to consider is that although Manus is currently classified as a closed-source, proprietary engine (just like ChatGPT and Operator), it’s built on open-source technology. The LLM “mind” is built on Anthropic’s Claude LLM, as well as Alibaba’s Qwen. The developers have said that they plan to open-source Manus itself – or parts of it – in the near future.
Is This AGI?
No, it doesn’t qualify as AGI – as the developers themselves state in their video, it’s a “glimpse” of AGI – artificial general intelligence.
AGI is best thought of as an eventual goal of developing AI that can learn to carry out just about any task, much like a human can. It uses the tools it’s given or, if it doesn’t have anything suitable, can work out how to find the tools it needs, or even design and use new tools.
It seems best to think of Manus as a collection of tools and models operating under agentic control, rather than one integrated software entity (like ChatGPT4o, for example).
It is, however, a clear sign that AI is evolving as fast, or perhaps faster, as expected. In a little over two years, since ChatGPT was released, we have seen AIs move from reactive, question-and-response systems to agents capable of actively managing digital and real-world tasks.
Geopolitical Implications
In recent years, we have seen AI become a geopolitical battleground, primarily between the U.S. and China.
The leadership of both nations has made it clear they recognize its enormous potential to shape the future due to its impact on everything from medical research to the economy to warfare.
Historically, it’s fair to say, U.S. companies have dominated the market, with OpenAI, Google, Microsoft and so on, leading on research and bringing products to market.
China, where the government plays a more hands-on role, has made significant strides to close the gap, however, first with DeepSeek and now potentially with Manus.
DeepSeek caused a crash in the price of U.S. tech stocks when it appeared that a Chinese model was able to closely match the performance of top-ranked U.S. models, at a fraction of the cost.
If Manus proves to be as useful as its developers predict, it could signal a further shift in AI dominance, potentially prompting Western players to rethink their strategies.
Everyday Agents
So, the race is on to create what I feel will become the equivalent of the computer operating system for the AI era. Essentially, agentic AI that enables anyone to do things with machines that previously would have taken expert-level knowledge and skills.
Because of this, AI and automation will become increasingly ingrained in daily life as we use them to assist in many everyday and workplace tasks.
This, in turn, will lead to a fundamental shift in the nature of the relationship between humans and technology as, for the first time in history, it becomes capable of acting autonomously on our behalf.
