Built on the vDisk cloud desktop central control platform, AI capabilities are embedded directly into the computer lab, with no need to purchase GPU servers,Cut costs by over 95%, giving every student ready-to-use AI tools.
vDisk AI teaching-environment deployment solution是Shanghai Chengcheng Information Technology Co., Ltd.an all-in-one AI teaching solution for university computer labs. The solution embeds AI large-model invocation capabilities vDisk Converged Cloud Management Platform; administrators can useImage MarketplaceDeploy images with built-in AI teaching components for simpler bulk rollout. The platform provides OpenAI API compatibleWith the proxy, VS Code, Cursor, various plugins and in-house software can connect to domestic large models via a unified protocol. Students can learn and program right inside the cloud desktop, and AI vision technology enables automatic classroom inspection and intelligent control of IoT devices.AI Teaching SpaceSupporting vDisk network disk (cloud drive): mounted as a virtual disk, with AI outputs and user directories stored on the server side, supporting session continuation across classes and Token savings (see the "Use Cases" below for details).
at full power DeepSeek-R1 (671B parameters)For example: the 8 H100 80GB GPUs required to run it cost approximately 1.5 million RMB; plus complete hardware,Total cost: 2–3 million RMB per set. Beyond hardware, you also face:
The logic of the vDisk solution:Rather than building proprietary models, it uniformly integrates the market's finest commercial AI APIs and, through the vDisk platform, delivers cost control, permission allocation, and usage analytics—letting schools enjoy top-tier AI capabilities at extremely low cost.
Embed large AI models directly into the cloud desktop; administrators centrally control usage and costs.Image MarketplaceChoose or deploy images with AI teaching components and clients pre-integrated, enabling batch deployment and version consistency,Make launching the AI teaching environment simpler. The platform provides OpenAI API compatiblethe proxy endpoint, any software that supports custom APIs and can be configured in the OpenAI style (such as VS Code, Cursor, all kinds of AI plugins, and in-house programs) can connect, and the backend then forwards the requests to a domestic large model.Bundled with vDisk network disk: virtual disk mounting, directory redirection and real-time server storage, all in the same closed loop as the AI teaching space.
vDisk Image MarketplaceProvides image templates with the client, teaching environment and common components already integrated; administrators only need to select or push an image to deploy it in bulk to the lab, reducing per-machine installation and version drift,The AI teaching environment and network-disk policies can be delivered together, for worry-free operations.
Platform External Offerings OpenAI API compatiblethe proxy address (including common paths such as chat completions). Teachers and students only need to enter the unified gateway and key in the settings of the tools they use,Call it using the OpenAI protocol; there's no need to integrate a private SDK separately for each large model — VS Code / Cursor / domestic AI plugins / in-house applications can all reuse the same approach.
Use Cases · AI Teaching Space · Network Disk (Cloud Drive)
Token-saving and resumable:Data generated by running the AI model is stored inNetwork Disk; the next class can pick up from the lastResume from Breakpoint, without re-entering large blocks of context, therebySignificantly save on Token consumption.
Product form (important):它NotWeb Cloud Storage,Not FTP、也Notnot an ordinary Windows shared folder, but rather one integrated withSame-level logic as the system boot drive (C:)的Virtual Disk, shown as a local disk in "This PC" (e.g. D:, NTFS), mounted and managed by the vDisk client.
Mounting & Writing:UsageUsername + PasswordMount a network disk to access storage directly over the campus network; dataReal-time write-back to server, what you see locally is what is stored.
Auto Redirect:Once mounted, the system automaticallyDesktop, My Documents, AppData and other key user directoriesRedirectionto that network disk. The desktop and files a student sees after logging in are their personal data environment on the server side.
Client Management:vDisk provides "Network Disk" management interface, where you can view drive letters and space usage, and performMount disk / unmount diskand other operations (consistent with the actual lab deployment interface).
Security Boundary:DataAvailable only on teaching endpoints where the vDisk software is deployedsaved and viewed centrally, preventing arbitrary copying and leakage; typicallyNo longer dependent on external content auditingis enough to meet classroom teaching management needs.
Note: "network drive" here refers to vDisk Network Disk(Computer-lab virtual disk mounting) — this is a different category of product from third-party internet "file-download cloud drives."
Virtual disk mounting (e.g. D:), not a web page / FTP / ordinary shared folder; Desktop, My Documents and AppData are redirected to the server and saved in real time. AI outputs saved to disk allow lessons to be resumed seamlessly, saving tokens. Mounting requires an account and password, and access is restricted to vDisk terminals only.
Select or deploy images from the image marketplace that already bundle the AI teaching client, teaching environment, and common dependencies, covering lab terminals in bulk to avoid per-machine installs and version mismatches — getting AI teaching live faster.
The platform provides an OpenAI API-compatible proxy endpoint; VS Code, Cursor, JetBrains-family plugins, domestic AI coding assistants, in-house applications, and more can connect simply by configuring the gateway and key in the OpenAI manner, with the platform uniformly forwarding requests to domestic LLMs.
The AI client can be placed in the student's desktop directory; combined with unified account authorization, students can use the capabilities of models such as DeepSeek, ERNIE Bot, Tongyi Qianwen, Zhipu GLM, Doubao, iFlytek Spark, Tencent Hunyuan and Kimi right after boot.
Track each student's Token consumption in real time; administrators can set a per-person usage cap to prevent abuse and cost overruns, keeping operating costs transparent and controllable.
After a class ends, AI conversation records are archived automatically, and the previous context is loaded directly for the next session, saving Tokens while making it easy for teachers to review the learning process.
The AI knowledge base can be customized to the institution's disciplinary characteristics—such as the "Shanghai Lixin University of Accounting and Finance program overview"—so that when students ask questions, the system automatically answers with the relevant disciplinary context.
The backend lets you switch the connected domestic model at any time (DeepSeek, ERNIE Bot, Tongyi Qianwen, Zhipu GLM, Doubao, iFlytek Spark, Tencent Hunyuan, Kimi, etc.), while the front end still uses the same OpenAI-compatible entry point, avoiding lock-in to any single vendor.
By feeding existing surveillance cameras into AI visual analysis, you get automated lab inspection, safety alerts and schedule-linked automation—with no extra hardware.
During class periods, cameras monitor the classroom and record class/no-class progress; during non-class periods, recording is triggered and violations are retained as evidence.
During breaks between classes, AI detection checks for lingering personnel, burning materials, or gas leaks. Anomalies are instantly pushed to the WeChat Official Account and the management backend, with 24/7 unattended operation.
Automated inspection of connected cameras and IoT devices, with inspection status shown in real time on the large-screen dashboard and automatic alerts for abnormal devices.
During class-free periods, security inspections are triggered automatically, focusing on left-on devices, lingering personnel and over-frequency checks, with devices shut down automatically when thresholds are exceeded.
Based on the class schedule, automatically shut down lab equipment (computers, air conditioning, lighting, access control) after class; pre-warm the environment ahead of new sessions and send operation notifications.
Before class begins, the system completes everything automatically per the timetable: unlock access control → turn on lights → turn on air conditioning → power on and launch the cloud desktop, with no manual operation required.
Surveillance, access control, air conditioning and lighting are all integrated into a single vDisk platform, with dual control via PC and WeChat Mini Program—no switching between multiple systems.
The same platform supports both a PC console and a WeChat Mini Program, with fully equivalent functionality on both. On site, there is no need to find a computer to open the console—everything can be done from a phone.
Cloud desktop management, monitoring, access control, AC and lighting control, timetable setup, and log viewing.
95% feature parity, managed anytime and anywhere, free of on- or off-campus network restrictions.
No need to equip every terminal with a GPU — a single server provides the AI runtime environment for the entire lab. Built on the IDV local-cache architecture, the AI environment starts in seconds with no lag.
Traditional self-built solution: each lab room requires 1 GPU server (2–3 million RMB)
vDisk cloud lab: one ordinary server manages AI images for multiple labs,Hardware costs approach zero
Embedded in the vDisk system, it automatically adapts to all kinds of endpoint hardware with no manual driver installation, and is compatible with the full IDV/VOI cloud desktop lineup.
The AI runtime is fully cached on the local terminal and executes from the local SSD based on the IDV architecture, launching 3-5 times faster than network boot and remaining unaffected by network jitter.
A ready-made AI environment image is provided with no complex configuration required; after the image is deployed from the vDisk admin console, students enter the AI experiment environment as soon as they power on.
Real case: At the multi-campus Shanghai Lixin University of Accounting and Finance, a single server manages all campuses, the AI environment images for each campus are maintained centrally, and the networks across campuses are interconnected.
Students can use VSCode with domestic AI coding-assistant plugins directly within the cloud desktop, or perform AI-assisted programming through the platform's built-in AI client, with no need to configure an API key separately.
Offer an AI general-education course for non-computer-science students, who experience large models through a unified AI chat interface and access AI technology with zero barriers.
AI inspection replaces manual on-duty monitoring, automatically detecting hazards when the lab is unattended after class and working with access control to lock down automatically, safeguarding property.
Lab administrators no longer need to power machines on and off manually each day; based on the timetable, the system fully automates the workflow: power on → authorize → class ends → power off → lock door.
Ideal for institutions with multiple campuses: a single management server manages the AI environments of several campuses at once, reducing operations staffing costs.
For institutions that cannot afford million-yuan GPU purchases, AI teaching can be delivered at extremely low cost via API integration, meeting the needs of educational informatization.
Contact us for a customized solution and quote tailored to your school's lab environment.