DSH Open-Source Solution Tested: Seamless Remote Control Across Phone, PC, and Cloud

An AI Workflow That Connects Three Devices

As AI tools become increasingly mainstream, enabling seamless collaboration between smartphones, computers, and cloud servers has become a major pain point for power users. Bilibili creator Jack recently demonstrated and open-sourced a solution called DSH — a three-way interconnection framework that bridges mobile, local PC, and cloud server environments, complete with cross-device remote control capabilities. This article breaks down the core architecture and practical value of the system based on his demo.

The strength of this solution lies not in any single feature, but in how it chains together AI capabilities scattered across different devices into a unified workbench — connected through a consistent interface and synchronization mechanism. For developers and AI practitioners who frequently switch between multiple devices, this kind of architecture can significantly reduce operational friction.

Live Demo: Trigger a Task on Phone, Receive It on PC in Real Time

In the demo, Jack selected an image and a document (referred to as a "bamboo draft") on his phone, used DSH to have the AI describe both files, and had the results sent to his PC's download folder. The key detail here: the phone was connected to DSH running on the cloud server, not the local instance.

After hitting send, the corresponding operation appeared in real time on the cloud DSH session UI — the interface remained consistent across devices. Meanwhile, the locally running DSH session stayed clean, clearly distinguishing which environment each operation belonged to.

The value of this UI sync mechanism is clear: no matter which device initiates an action, users always see a consistent context — eliminating the "information gap" that typically plagues multi-device workflows.

Multimodal Understanding with End-to-End File Delivery

The second highlight of the demo was the complete end-to-end loop. The AI successfully described both the image content and the text content of the document, and the generated files — including the document text and image — actually appeared on the local PC.

This workflow validates three connected layers: data uploaded from the phone, computation handled in the cloud, and results delivered to the local PC. Three physically separate devices formed a collaborative pipeline in a single operation. For users who need to kick off tasks on mobile and receive results at a desktop, this cross-device workflow offers real practical value.

Dual-Service Architecture: Mutual Failover by Design

What makes this solution especially noteworthy is its architectural design. Jack configured two services in the system: one to control his local PC, and another to control the remote server.

This dual-service design creates an elegant disaster recovery capability:

• If the DSH instance on the cloud server goes down, the user can use their phone to control the local DSH and bring the cloud instance back up;
• Conversely, if the local DSH goes down, the cloud DSH can be used to restart the local service.

In other words, the two DSH instances serve as mutual operational backups — forming a kind of decentralized, self-healing mechanism. And all of this can be done from a smartphone, meaning you can handle service failures remotely even when you're away from your computer.

Fully Open Source, with Support for Agent-Based Auto-Deployment

According to Jack, the entire interconnection and mobile remote control workflow has been open-sourced on GitHub. What's more, he mentions that users can let their own AI Agent automatically install both tools — which itself reflects a growing trend in AI workbenches: even the deployment step can be handed off to an agent, further lowering the barrier to entry.

For developers looking to build a personal AI workbench, this open-source project offers a concrete reference implementation. Its value lies not just in the features themselves, but in demonstrating an architectural pattern centered on "multi-device, multi-instance, remotely operable" infrastructure.

Summary: The Core Value of DSH's Three-Way Integration

At its heart, this DSH three-way integration solution is an engineering effort built around one goal: "control my AI capabilities from anywhere, at any time." Its three core values can be summarized as:

1. State Synchronization — Consistent UI across all devices, eliminating information gaps;
2. Capability Collaboration — Phone, PC, and cloud each play their role while remaining interconnected;
3. Operational Resilience — Dual instances serve as mutual backups, with remote recovery possible from a phone.

It's worth noting that this article is based on a single source (Bilibili creator Jack's demo), and the actual stability, performance, and security of the solution still need to be validated by more users in real-world environments. That said, as an architectural blueprint for building a personal AI workbench, it undoubtedly offers valuable inspiration for the community. Interested readers can visit his GitHub repository for deployment details.