Alibaba's Qoder allows sending coding tasks from one desktop client to another device without using SSH

Tencent's artificial intelligence agent platform, WorkBuddy, has released a remote control feature supported by the Tencent Edge One team. This new capability allows users to connect to a computer logged into the same account using a mobile application. Users can view the desktop in real-time and intervene in operations if necessary.

Unlike traditional remote desktop tools, which primarily focus on device-to-device access, the WorkBuddy feature is specifically designed for AI agent workflows. Users gain the ability to monitor the agent's actions in real time and intervene if tasks deviate from course or require manual confirmation.

To use this feature, the mobile application must be updated to version 2.7.0 or later, as well as the macOS desktop client to version 5.9.3 or higher. After connecting, the computer screen is displayed in a pop-up window on the app's home page or at the top of the dialogue, allowing monitoring of the agent without leaving the current chat.

If an agent encounters an issue while performing a task, users can intervene directly from their phones. For example, if an account is not authorized or the connector authorization has expired, users can attempt to resolve the issue on mobile devices instead of returning to their computers.

For manual control, users can press the full-screen mode button, which switches the phone interface to landscape orientation and activates the ability to control the desktop via touch input. The interface provides three options: Control, Keyboard, and Mouse, which support actions such as switching windows, returning to the desktop, and text input. A virtual mouse is also available for more precise clicking and scrolling in small areas of the screen.

Regarding security, Tencent states that the feature limits each device to one authenticated mobile device simultaneously. Screen data and control commands are transmitted over an encrypted channel. The platform does not record user input or store video recordings of remote control sessions. Network traffic generated by remote control is deducted from the device's data usage and does not consume WorkBuddy AI agent token quota.

To get started, users need to enable the remote connection option for devices logged into the same account in WorkBuddy settings, install the necessary remote control component, and grant screen recording and accessibility permissions. Once configured, users can connect remotely while the computer remains powered on and WorkBuddy is online.

In practice, this feature helps solve a common problem in AI agent workflows—the need for human intervention during task execution. AI agents are already capable of performing multi-step tasks, such as document organization and data collection, but they can still encounter authorization requirements, permission requests, or operational errors. Remote control gives users the ability to monitor progress and make adjustments from their phones without being tied to their desks.

However, this feature may offer more than just a way to control computers from a phone. It could potentially introduce a more flexible model of human-AI collaboration, where AI handles routine work, and users intervene to make decisions and correct errors at critical moments. Rather than striving for completely autonomous execution at all costs, maintaining human oversight might be a more practical approach for AI productivity tools at the current stage of development.

Nevertheless, a smartphone cannot fully replicate the computer experience. Remote access may be convenient for simple tasks, such as approving permissions or switching windows, but its effectiveness in more complex workflows involving data processing and multiple windows remains to be seen. The feature is currently available for trial use by all users. Tencent reports that future versions will include more enterprise-oriented capabilities, including role-based permission management and audit logs.

Alibaba, in collaboration with Alibaba Token Hub, has introduced the λ0 model. This foundational model is designed to train humanoid robots in coordinated movements and manipulations by utilizing demonstrations shown by humans.

The system learns from first-person video recordings that are synchronized with body and arm movements. This enables robots to perform more complex tasks, combining walking, posture correction, and two-handed operations.

Researchers created the HumanVerse dataset, which includes 500 hours of recordings of people performing various actions in real-world conditions. This approach reduces the need to collect demonstrations through robot teleoperation.

During testing, the team demonstrated that λ0 achieved 62.5% success rate in tasks related to locomotion and manipulation in the real world. The project also provided researchers with configuration code for data collection equipment and data samples.

DAMO Academy from Alibaba unveiled the XuanTie E908A automotive processor IP at the RISC-V summit in Shenzhen, China, on October 19. This processor IP is designed for safety-related systems, domain controllers, and automotive gateways, and complies with automotive functional safety requirements ASIL-B and ASIL-D, according to DAMO Academy.

Furthermore, the company disclosed a processor roadmap intended for autonomous driving and embodied artificial intelligence. The planned automotive processor C950A is expected to deliver five times the performance of current RISC-V products for the automotive industry, with its release anticipated by the end of 2027.

DAMO Academy reported that the cumulative volume of chips utilizing the XuanTie processor IP has exceeded 6 billion units as the company expands its application from established areas into high-performance computing markets.