FAQ

Please rest assured that ordinary computers with lower configurations can run this software smoothly. The actual size of the software core program is only about 200MB. The reason why the download package size reaches 1.5G is mainly because it includes the following additional resources to achieve an efficient "out-of-the-box" experience:

• 1. AI drivers adapted to different graphics cards

  In order to be compatible with various graphics cards (such as NVIDIA, AMD, etc.) to achieve local AI acceleration, multiple AI driver environments are integrated into the installation package. During actual operation, the system will only load one of the adapted drivers based on your computer hardware, and the other unsuitable drivers will not be loaded at all and will not occupy your system's running resources.

• 2. Built-in local AI model (about 600MB)

  To allow you to start using it immediately after installation without the need for cumbersome network downloads, we have pre-built a lightweight and efficient local AI model. The model is primarily loaded into graphics/memory when running AI inference. If you download and enable other new models later, the built-in models will be automatically deactivated and no longer occupy any runtime resources.

2.0 directly supports the latest data structure and does not support data migration from older versions. Users who want to upgrade please refer to the following solutions:

• Version 1.1.0 and above

  Can be directly upgraded to 2.x

• Users below 1.1.0

  Please upgrade to any version 1.1.0~1.3.2 first, and then upgrade to 2.x

Yinghe Smart Folder supports local and cloud APIs, performs local file analysis, and is a professional tool for automatically organizing files to protect your privacy. It has launched Qwen 3.5 and removed the restricted version, making it more intelligent and powerful.

Yinghe Smart Folder adopts multiple privacy protection mechanisms:

• Runs entirely locally

  All AI analysis is done locally on your computer and the files never leave your device

• Extreme speed mode

  Because AI analysis data is time-consuming, in extreme speed mode, the software will upload your file analysis data (not the file itself) to a cloud server for sharing (detailed below), so that others can speed up the analysis when they download the same file from public resources on the Internet, realizing the ideal practice of everyone for me and me for everyone.

• privacy mode

  Privacy mode is the opposite of extreme speed mode. The software will not upload your file analysis data to the cloud server. You cannot enjoy cloud acceleration in this mode, but it can ensure complete privacy of your files.

• Open source auditable

  The business logic is completely open source and anyone can review the code to ensure there are no backdoors

• Available offline

  It fully supports offline use and can complete file analysis without being connected to the Internet. It is recommended to use the local mode Qwen3.5 series community version model, which can avoid the online model using your data for AI training.

Yinghe Smart Folder is the personal work of a geek idealist. It is completely free and the core code is open source. But it is not completely public welfare. We must ensure the continued development of the software, so:

• Free forever

  The software will continue to develop more powerful functions and ensure that these functions are free for most users, and only charge for some users with high requirements, such as those who have high-frequency use requirements for a certain function in the future.

• Sustainable operation

  On the other hand, project operations are maintained through enterprise customization services, technical consulting, etc.

• Thanks for the support

  If you like this software, please feel free to like and collect it on Bilibili videos, Zhihu answers, WeChat public accounts, etc., and recommend it to your friends. This will greatly help the development of the project.

Yinghe Smart Folder always takes privacy protection as its core design principle, insists on giving priority to data localization, and effectively protects user privacy and security. Its core technical principles are logically consistent with the common hash (Hash) mechanism in the industry.

• Simply put, the system will use an encryption algorithm to generate a unique characteristic identifier (length 16-128 characters) for each file, which is equivalent to the unique ID of the file.

• The only prerequisite for generating this ID is that the user holds the file locally and completes the algorithm calculation. The application will not disclose the file ID to any third party.Only with this ID can the analysis data corresponding to the ID be requested from the server, and there is no other way to obtain it.

• Even under extreme assumptions (the server is breached), third parties can only obtain file analysis data and corresponding IDs, and the ID algorithms of different platforms are completely independent and have no connection with the file IDs of platforms such as Baidu Netdisk, making it impossible to match original files across platforms.

• At the same time, the server only stores analysis data: including smart file names, tags, content summaries, ratings and reasons, metadata, etc., which is what you can see in the properties panel on the right. The file itself will never be uploaded or stored.

Yinghe Smart Folder supports the following commonly used shortcut keys to help you manage files efficiently:

• Backspace

  Trigger back navigation

• Ctrl+A / Cmd+A

  Select all files in current list

• Escape

  Clear all currently selected items

• Alt + ↑

  Navigate to higher level directories

• Alt + ←

  Back navigation

• Alt + →

  Forward navigation

If you are not satisfied with the smart file names generated by AI, you can optimize them in the following two ways:

• Method 1: Regenerate smart file name rules

  Select the directory where the file is located and click "Reanalyze" at the bottom of the right sidebar. AI will rescan the directory and generate a new naming strategy.For example, you might see the AI ​​generate a format like "literature_serial_content_description".If you are not satisfied, you can reanalyze it multiple times until you are satisfied. It is recommended to switch to a smarter model to obtain a more accurate naming effect.

• Method 2: Modify the AI ​​prompt word

  In the "AI Analysis Settings" on the software settings page, you can directly modify the prompt words (Prompt) for label generation and intelligent file name generation to customize the naming logic of AI.