Manus and Lin Junyang have both returned.

On August 12, two star projects and figures that once attracted wide attention in the AI circle announced their return.

On August 11 local time in the United States, AI startup Manus notified its users that the company will "soon resume operations as an independent company". As part of the separation from Meta, data generated by some users on and after December 29, 2025 will be deleted in accordance with regulatory requirements.

According to the company's notice, affected users need to complete data backup before 7:59 a.m. Singapore Time on August 23. Between August 23 and August 24, relevant data will be deleted and affected accounts will be temporarily inaccessible. Starting from 8 a.m. Singapore Time on August 25, users can restore their previously backed up data and reuse the service.

Manus stated that this adjustment is not caused by data breaches or security incidents, but a measure taken to meet the regulatory requirements of specific jurisdictions during the process of the company resuming independent operations. Unaffected users do not need to take any action and can continue to use the service normally.

A few hours later, another important return news came from Lin Junyang. The former technical lead of Alibaba's Qwen project announced the establishment of AI lab Pragmatik (p7k) Labs in Shanghai, focusing on research on next-generation agents in the digital and physical worlds.

Both are agent-related projects, concerning the return of a company and the new journey of a technical leader respectively, but both point to the same change: AI competition is extending from model capability itself to how models can call tools, adapt to the environment and complete practical tasks.

01

The Meta Acquisition Enters the Termination Phase

The deal between Manus and Meta dates back to December 29, 2025.

On that day, Meta announced plans to acquire Manus. Founded in 2022, the company initially developed in China before moving its headquarters to Singapore, focusing on developing AI agent products.

The two parties did not announce the transaction amount at that time, but previous reports showed that the transaction was valued at about 2 billion U.S. dollars, with a potential total value of up to 2.5 billion U.S. dollars including employee retention arrangements.

After the acquisition was completed, Meta planned to apply Manus's AI agent technology to its own consumer and enterprise products to enhance its AI business capabilities.

However, the transaction subsequently drew regulatory attention.

On April 28, CCTV News reported that the highly concerned Manus acquisition case reached a final conclusion: the Office of the Working Mechanism for Security Review of Foreign Investment in China (National Development and Reform Commission) made a decision to prohibit the foreign investment in the Manus project in accordance with laws and regulations, and required the parties to cancel the acquisition transaction.

Since then, Manus and Meta have started to promote operational separation and stopped data sharing between the two parties.

According to previous information, the two parties completed the operational separation in May. At present, Manus is completing the final stage of data processing and preparations for independent operation.

This data adjustment mainly involves data generated on December 29, 2025 and afterwards.

Manus stated that the relevant data deletion is to meet regulatory requirements. The company has launched data backup and recovery tools to help affected users save their task records.

Affected users can perform multiple backups during the backup window. If users continue to generate new task data after completing the first backup, they need to back up again to ensure that the latest data is saved.

Manus stated that no fees will be charged to affected users during the backup period. After restoring their accounts, the company will also offer a return bonus.

For users registered with Apple ID or Facebook account, since Manus may not have the corresponding email address, the company reminds users to pay attention to in-app notifications.

02

Manus Founding Team and Product Roadmap

The company behind Manus is Butterfly Effect, founded in 2022.

Founder Xiao Hong has long been engaged in enterprise software development. According to public information, he studied software engineering at Huazhong University of Science and Technology, founded Wuhan Nightingale Technology after graduation in 2015, whose WeChat tools "Yiban Assistant" and "Weiban Assistant" serve more than 2 million business users.

Butterfly Effect later launched Monica, an AI assistant product integrating capabilities of multiple language models. Manus further extended the AI capabilities to the field of task execution.

Manus is positioned as a general-purpose AI agent platform, where users can use natural language instructions to let the system complete tasks such as information search, report writing, file processing, and code development.

Co-founder and chief scientist Ji Yichao previously developed an iPhone browser and founded Peak Labs, focusing on information extraction and search technology. Co-founder Zhang Tao previously was in charge of product-related work at ByteDance and Light Years Beyond.

On March 6, 2025, Manus was officially released to the public.

03

Rapid Growth Before Acquisition, Future Equity Structure Remains Undecided

Before being acquired by Meta, Manus had completed multiple rounds of financing and achieved rapid growth.

Earlier in 2025, the company secured $75 million in financing from Benchmark, with a reported post-money valuation of around $500 million.

In December 2025, Manus stated that eight months after the product launch, its annualized recurring revenue exceeded 100 million U.S. dollars, and the total revenue run rate reached 125 million U.S. dollars.

The company also disclosed that the platform has processed more than 147 trillion tokens in total and created more than 80 million virtual computers.

When Meta acquired Manus, the two parties hoped to further expand their AI agent business. However, with the termination of the transaction, Manus's future equity structure and financing arrangements have not yet been announced.

According to previous reports, during the process of terminating the transaction, the founding team explored raising about 1 billion U.S. dollars to buy back the company's shares.

Some early investors also participated in relevant discussions, including institutions such as ZhenFund and Hony Capital.

Manus stated that after resuming independent operations, the company will continue to provide services to global users and plans to launch new product features. It will further enhance its AI agent capabilities in the future, but no specific product roadmap has been announced.

In terms of data storage, Manus stated that after resuming independent operations, user data will continue to be stored in the United States and Singapore.

04

From Training Models to Building Agents, Lin Junyang Embarks on New Exploration

The resumption of Manus's independent operations means that this startup focused on general agents will regain the space to independently adjust its product and financing arrangements. On the same day, Lin Junyang, another important technical leader who once led the R&D of domestic large models, also set his entrepreneurial direction on agents.

Lin Junyang announced his resignation from Alibaba on March 3, 2026. Previously, he had long been involved in large model R&D, and focused on reasoning capabilities, agent training, and the interaction between models and the environment during the development of the Qwen series of models.

Moving from reasoning models to agent systems means that the focus of AI competition is changing. Future technological breakthroughs will not only come from the models themselves, but also from how models are combined with tools, environments and real-world tasks.

Lin Junyang focuses his new research direction on agent systems. The name Pragmatik Labs comes from "Pragmatics" in linguistics. Lin Junyang explained that he first learned linguistics because a friend recommended pragmatics to him, and then he turned to computational linguistics and natural language processing. "Pragmatik" represents returning to the place where things happen, and also reflects the team's understanding of pragmatism, that is, AGI ultimately needs to solve practical problems in the real world.

Pragmatik Labs focuses its research on two directions: digital agents and physical agents.

In the digital field, the company hopes to build general agents for knowledge work, enterprise operations and industry processes, so that AI can handle more complex work tasks. In the physical field, the company explores embodied intelligent systems that can enter the real environment, adapt to changes and perform long-term tasks.

Pragmatik Labs introduced on its official website that next-generation agents need to have capabilities of reasoning, tool calling, learning from feedback and coordinating actions. Different from traditional models that mainly complete information generation, agents need to run continuously in the environment, adjust strategies according to results, and achieve long-term goals.

This direction is consistent with Lin Junyang's previously proposed view of "training models → training agents". In his view, reasoning models solve the problem of how to make the model perform more effective internal calculations before answering, while agent systems focus on how to make the model take actions in the real environment. Future AI systems need to deal with problems such as tool selection, task planning, environmental feedback, failure correction, and multi-round task collaboration.

Pragmatik Labs is currently making layouts around this direction, hoping to extend AI capabilities further from digital information processing to real-world tasks.

This article is from the WeChat Official Account "Tencent Tech", Author: Joanne, Editor: Su Yang, published with authorization from 36Kr.