laitimes

Another domestic dark horse after DeepSeek? The "world's first" general AI Agent swiped the screen overnight, and the whole network was asking for an invitation code

Author|Dongmei

After DeepSeek, another AI product detonated the tech circle last night.

A team called Manus in China has released an AI Agent product, which is introduced on the official website of Manus as follows: Manus is a universal AI Agent, which connects thought and action: it not only thinks, but also provides results. Manus is good at handling a variety of tasks at work and in life and can do everything while you are resting.

1. The first general-purpose AIAgent detonated the technology circle

It is said that this is the world's first truly general-purpose AI agent.

According to the Manus team, unlike traditional AI assistants, Manus is a truly autonomous AI agent, which can not only provide suggestions or answers, but also directly deliver complete task results.

Manus Experience Application Link: https://manus.im/invitation

Specifically, what can it do?

In the 4-minute video presentation released on the official website, the resume screening work is first shown. Manus was sent a compressed file of 10 resumes, and Manus was able to sift through resumes as required like a human HR.

Manus unzips the files, then goes through each resume one by one and extracts key information. It supports asynchronous processing, so you can shut down your PC while a task is in progress and it will notify you when it's done. In the meantime, you can also send new commands to Manus at any time. Next, go ahead and upload 5 resumes. After a careful analysis of all 15 resumes, Manus provides a ranking recommendation, along with candidate profiles and evaluation criteria for reference.

Manus can even be asked to generate spreadsheets based on the user's usage habits.

Manus has the knowledge and memory skills, so the next time it performs a similar task, it delivers the results directly in spreadsheet.

In the second demo video, Manus was asked to choose a number of livable neighborhoods, which would take into account the family's income and the needs of the children in the family, and the location should be in New York, where the conditions should be safe and the crime rate should be low.

As you can see, there are several smaller requirements in this requirement. When tackling such complex tasks, Manus can easily break it down into multiple steps and create a detailed to-do list.

  • Search and read articles about New York's safest neighborhoods.
  • Research the situation in New York's secondary schools.
  • Write a Python program to calculate a budget.
  • Based on your budget, you can filter for suitable listings on real estate websites.
  • Consolidate all the information, write detailed reports and organize relevant materials

In the third presentation, Manus is a professional equity analyst tasked with analyzing the correlation between stock prices of Nvidia, Marvell Technology, and TSMC over the past three years.

It accesses authoritative financial data sources through API interfaces, ensuring the accuracy and timeliness of data. Once the data was validated, Manus set out to write code that would be used to drill down into the stock price movements of the three companies and generate intuitive visualizations to reveal correlations between them.

After an in-depth analysis and visualization of data on the stock prices of Nvidia, Marvell and TSMC, Manus showed its versatility by translating these analyses into an interactive website. With explicit authorization from the user, Manus deployed the insightful website to the cloud and made it accessible online. Users will be given a unique link that can be easily shared with peers, friends, or anyone interested in this analysis.

This process not only demonstrates Manus' strong capabilities in data processing and analysis, but also demonstrates its expertise in the financial sector. Through Manus' analysis, investors can better understand the interconnectedness of the stock price fluctuations of these three tech giants, so as to make more informed investment decisions.

With so many complex tasks, how does the agent perform in benchmarks?

According to the official website, in the GAIA benchmark, which evaluates the ability of a general-purpose AI assistant to solve real-world problems, Manus achieved a new state-of-the-art (SOTA) performance across all three difficulty levels.

According to reports, in addition to benchmarking, Manus also solves real-world problems on platforms such as Upwork and Fiverr, and has a strong performance in the Kaggle competition.

So, what is the origin of such an excellent product?

According to public information on the Internet, Xiao Hong, the founder behind Manus AI, is a 2015 alumnus of Huazhong University of Science and Technology majoring in software engineering.

Xiao Hong is a serial entrepreneur. Xiao Hong once was a guest on a talk show, in which he provided a mentality of an entrepreneur in the waves: "The world is not linear extrapolation, and you must make yourself an important variable in the game." ”

After graduating, in 2015, Xiao Hong founded Nightingale Technology, launched "Yiban Assistant" and "WeCompanion Assistant", served more than 2 million B-end users, and received investment from Tencent and Zhen Fund.

Xiao Hong is also the initiator of another star AI product, Monica.

Monica is an all-in-one AI assistant tool in the form of a browser plug-in, PC/Mac, and mobile platform. Monica integrates with the current mainstream large models, and users can chat, translate, handle copywriting, and create drawings with Monica anytime and anywhere.

Currently, the Monica plugin has more than 1.5 million users worldwide.

Xiao Hong shared the reasons why they chose to start with the product form of browser plug-ins at a technical conference:

"Overseas, browser plug-ins are a relatively mainstream product form. For example, Grammarly, with annual revenue of hundreds of millions of dollars, may have a browser plug-in user base of hundreds of millions.

Our core consideration is, how can AI as a capability be adapted to a variety of scenarios? Browser plug-ins happen to be the form of a product that we've found that does this very well.

Big companies have developed a lot of great AI capabilities, but many websites don't integrate these capabilities right away. When users use browsers, they really want to use AI capabilities, and browser plug-ins can bridge the gap in the experience of various websites from the user's perspective

There are also many successful cases of independent developers going overseas.

At the end of last year, just four days after the release of ChatGPT, wong2 (screen name), the head of the instant store "Jishito", released a plug-in "ChatGPT for Google" to integrate ChatGPT into the search engine, and it was open-sourced on the same day.

From having an idea, to starting to write code, to finally going live, WONG2 took less than 12 hours. More than two months after ChatGPT for Google was launched, it surpassed 1 million users and was immediately acquired by Monica, which is currently valued at tens of millions of dollars. ”

1 What do netizens think?

The release of Manus has generated a lot of attention in the tech world.

On X, http://Hallid.ai co-founder and former Weibo employee indigo posted:

“Manus 是首个通用 AI Agent,看起来真是太酷了。 ”

More people were shocked by Manus and asked for an invitation code......

So, what do you think of this product? Feel free to talk in the comment section!

Of course, it would be great if you could help me apply for an invitation code~

Reference Links:

https://manus.im/

https://www.xiaoyuzhoufm.com/episode/6586ca30157cd445eebb399c

https://www.jingpt.com/article/b2f05f89-6a30-4b68-832b-e1557b6e5c8f

https://x.com/cellinlab/status/1896433479916916799

Disclaimer: This article is compiled by InfoQ and does not represent the views of the platform and is not allowed to be reproduced without permission.

Today's good article recommends Tencent Ingot to modify the user agreement overnight! The "overlord" clause rushed to the hot list, who has the final say in your content?

"Plagiarism" code, is it the CTO's pot or the founder's pot?! It's been 3 days since this happened

Distributed systems programming has stalled?!

Father Curl: How did I sleep peacefully with 180,000 lines of C code on my pillow?

Read on