Manus, the “agent” AI platform that launched in preview last week, is creating more hype than the Taylor Swift concert.
Manus called “the most impressive AI tool I’ve ever tried,” a hugged face product head. AI policy researcher Deanball described Manus as “the most sophisticated computer using AI.” Manus’ official Discord Server has reportedly grown to over 138,000 members in just a few days, and is inviting Manus code.
However, it is not clear that the hype is justified.
Excellent https://t.co/tfev9qz1d0
– Jack (@jack) March 9, 2025
Manus was not developed entirely from scratch. According to reports on social media, the platform uses a combination of existing and fine-tuned AI models, including Claude of Humanity and Qwen from Alibaba to perform tasks such as drafting research reports and analyzing financial applications.
But on its website, Butterfly Effect, the Chinese company behind Manus, offers some wild examples of what the platform seems to be able to achieve, from purchasing real estate to programming video games.
In X’s viral video, Manus’ research lead, Yichao “Peak” Ji, implies that the platform is superior to Openai’s deep research and agent tools such as operators. Manus outweighs deep research into popular benchmarks for popular AI assistants called Gaia, and JI argued that it will investigate AI’s ability to use software to browse the web and perform work.
“(Manus) is not just another chatbot or workflow,” Ji said in the video. “It’s a completely autonomous agent that bridges the gap between concepts and execution (…) we see it as the next paradigm of human machine collaboration.”
However, some early users say Manus is not a panacea.
Alexander Doria, co-founder of AI startup Pleias, said in a post on X that he encountered an error message and an infinite loop while testing Manus. Other X users pointed out that Manus made mistakes in factual questions and did not consistently cite the work.
The in-depth study was completed within 15 minutes. Unfortunately, Manus AI failed after 50 minutes on Step 18/20! corty I was watching Manus’ output and played pretty well, and it seemed fantastic. However, running the same prompt again is a bit frustrating, as it takes too long. https://t.co/bgtmoi65cp
– Derya Unutmaz, MD (@deryatr_) March 8, 2025
My own experience with Manus was incredibly unpopular.
I asked the platform to handle what seemed like a rather simple request. About 10 minutes later, Manus crashed. On my second attempt, I found a menu item that met my criteria, but Manus was unable to complete the ordering process.
Manus likewise cried when he asked him to book a flight from New York to Japan. Given the instructions that I thought were not much room for ambiguity (“Find business class flights, prioritize prices and flexible dates”), the best manus was to link to the fare on some airline websites and airfare search engines like Kayak.
Hoping that the next few tasks might be attractive, I told Manus to book a table at a restaurant within walking distance. It failed a few minutes later. I then asked the platform to build a Naruto-style fighting game. It made an error in 30 minutes. That’s when I decided to throw the towel.
Honest opinions after trying Manus AI for the past 3 days, here are the good and bad things.
good:
– The research it does on the internet and the reports it generates are incredible.
– The ability to run scripts behind the scenes to perform tasks is impressive.
– plan…– March 9, 2025
A Manus spokesman sent the following statement via DM to TechCrunch:
“As a small team, our focus is to continue to improve our Manus and create AI agents that will actually help users solve problems (…) The main goal of the current closed beta is to stress-test different parts of the system and identify problems. We are deeply grateful for the valuable insights that everyone has shared.”
So, if Manusis has not reached its technical promise, why did it explode? Several factors contributed, including the exclusivity created by the lack of invitations.
Chinese media quickly promoted Manus as an AI breakthrough. Publication QQ News called it “pride in domestic products.” Meanwhile, social media AI influencers have spread misinformation about Manus’ abilities. The widely shared video showed the manus on the surface, a desktop program, and performed actions on multiple smartphone apps. Ji confirmed that the video wasn’t actually a Manus demo.
X’s other influential AI accounts sought to derive a comparison between Manus and Chinese AI company Deepseek. In reality, it is not necessarily ingrained. The Butterfly effect, unlike DeepSeek, did not develop an in-house model. And while Deepseek has made much of its technology openly available, Monica wasn’t – at least not at all yet.
To be fair to the effect of the butterfly, Manus is very quick to access. The company claims it is working to expand its computing power and fix issues, as reported. However, with the platform currently in existence, Manus looks like the case of hype ahead of innovation.
6:02 PM Pacific: Added a statement from Manus’ spokesman to correct the company misidentification behind Manus.