Frequently Asked Questions

AI FileSorter is a desktop app that helps organize messy folders using local or remote AI models. It focuses on practical cleanup tasks like categorization, smarter filenames, preview-first sorting, and privacy-aware processing.

Yes. AI File Sorter supports fully offline use with local models such as Gemma 3 4B IT, Gemma 1.1 7B, and Mistral 7B. It can also run local image analysis with supported visual backends, so file processing can stay on your own machine.

You only need an internet connection if you choose to use a remote model with your own OpenAI or Gemini API key.

Yes. If you prefer remote inference, you can bring your own OpenAI or Gemini API key and choose cloud-based models instead of local ones.

It can work with:

• general files and folders
• supported picture files
• supported document formats such as PDF, DOCX, XLSX, PPTX, ODT, ODS, ODP, and common text files
• supported audio and video files for metadata-based filename suggestions

Yes. AI FileSorter includes a preview-only dry run with a From/To table, so you can review the plan before confirming any moves or renames.

Yes. The app supports rename-only workflows and review flows with clear status labels, so you can use filename suggestions without reorganizing folder locations.

Yes. AI FileSorter includes persistent undo, which saves a plan file so you can revert the latest sort even after closing the dialog.

AI FileSorter is available for Windows, macOS, and Linux.

It is designed for privacy-aware file organization. If you use local models, your data stays on your machine. If you use remote models, you provide your own API key for the service you choose.

The best place to start is this guide: How to Organize PDFs and Office Documents with AI Without Uploading Them to the Cloud.

Yes. Open Settings -> Select LLM..., expand Downloads, and choose a different folder under Model storage.

If you already downloaded a model, you can move the existing model files into the new storage folder instead of downloading them again. Make sure the files keep their original names.

For image analysis models, move both:

• the main .gguf model file
• the matching mmproj file

You can also add a .gguf file from another location as a custom local model.

Skipped means AI File Sorter did not run that specific analysis step for that file.

For example:

• image analysis may be skipped if the visual model is unavailable, model startup fails, or the app falls back to filename-based sorting
• document analysis may be skipped if the file cannot be read as text or the analysis step fails

When this happens, AI File Sorter will usually continue with the safest available fallback, such as using the filename, cached results, or metadata. Check the progress messages for the specific reason.

Not yet. AI File Sorter can analyze text-based PDF files when the text is already embedded in the document, but scanned PDFs are usually image-only files and currently require OCR first.

OCR support for scanned PDFs is planned for a future version. For now, you can run OCR with another tool and save the file as a searchable PDF or text document before analyzing it with AI File Sorter.

Yes, partly. AI File Sorter can be guided with custom category/subcategory constraints, such as a whitelist, so suggestions stay closer to the categories you want.

You can use branching whitelists where you can set up your own categories and corresponding subcategories.

Not yet. AI File Sorter currently focuses on analyzing, categorizing, moving, and optionally renaming files. It does not currently detect or flag duplicate files.

Duplicate detection is planned for a future version.

If your computer restarts while AI File Sorter is analyzing files, the current run will not automatically resume. You can reopen the app and start the analysis again.

Depending on how far the previous run got and what was saved in the local cache, some results may be reused, but the app may need to analyze some files again.

Not yet. For now, review the results before confirming and manually deselect any files you don't want moved or renamed.

A file-type include/ignore feature is planned for a future version.

AI File Sorter does not move or rename files during analysis. The analysis step only prepares category and filename suggestions. When analysis finishes, a Review and Confirm window should appear.

Files are only moved or renamed after you approve the suggestions by clicking Confirm & Sort.

If the app was closed, cancelled, or the Review window disappeared before confirming, no files were changed. Run the analysis on the same folder again; previously saved categorization results may be reused so you can continue from the review step faster.

Image renaming requires image analysis to be enabled and the visual model files to be downloaded.

Open Settings -> Select LLM... and check the Image analysis models section. For the selected visual backend, both files must be downloaded:

• the model file
• the matching mmproj file

Then, on the main screen:

1. Enable Analyze picture files by content.
2. Expand the picture options.
3. Enable Offer to rename picture files.
4. Run analysis again.

For document renaming, the visual model is not required. Enable Analyze document files by content and Offer to rename document files instead.

The greyed-out checkboxes in the Review screen are rename-only toggles. They are only enabled when the run contains supported image/document files and the matching Offer to rename... option was enabled before analysis.

You can transfer local settings, categorization cache, custom whitelists, and learned review decisions. First, close AI File Sorter on both computers. Then copy the app settings folder from the old computer to the same location on the new one.

Settings folder locations:

• Windows: %APPDATA%\AIFileSorter\
• macOS: ~/Library/Application Support/AIFileSorter/
• Linux: ~/.config/AIFileSorter/

Copy these files if they exist:

• config.ini - app settings, selected AI provider/model, API keys, folder preferences
• categorization_results.db - cached categorization results
• user_learning.db - learned behavior from approved Review dialog decisions
• whitelists.ini - custom category/subcategory whitelists

Optional extras:

• review_history.sqlite - review/history records
• image_place_cache.db - cached photo GPS place lookups

After copying, start AI File Sorter on the new computer and check Settings. Some saved paths, such as source folders, destination folders, or custom local model paths, may need to be updated because Windows, macOS, and Linux use different folder formats.

Downloaded local AI models are stored separately. Copying them is optional, but can avoid re-downloading large files:

• Windows: usually %APPDATA%\aifilesorter\llms
• macOS: ~/Library/Application Support/aifilesorter/llms
• Linux: ~/.local/share/aifilesorter/llms

Starting with the next version, AI File Sorter will be able to recognize and work with existing folder hierarchies, including nested subfolders. This allows you to organize files using an established folder structure rather than having to recreate it from scratch.

This error means Windows tried to load AI File Sorter's CUDA acceleration backend, but a required NVIDIA CUDA runtime DLL was not available.

First, install the latest AI File Sorter release. Newer builds validate the CUDA backend before using it and should automatically fall back to Vulkan or CPU when CUDA is not usable.

You normally do not need to install the full CUDA Toolkit just to use AI File Sorter. If you do not specifically need CUDA acceleration, using Vulkan or CPU is fine.

Try these steps:

1. Update AI File Sorter to the latest version.
2. Update your NVIDIA graphics driver from NVIDIA's website. In this case, cublas64_13.dll refers to CUDA 13.x.
3. If you previously forced CUDA through an environment variable or advanced setting, remove that override and let the app choose automatically.
4. If the error still appears, choose Vulkan or CPU acceleration if available.

If the problem continues on the latest version, please report it using Contact -> Report issue.

AI File Sorter analyzes the files you select and proposes an organized result based on the sorting mode and settings you choose. Existing folders are not automatically treated as rules that must be preserved, but they can provide useful context depending on the workflow.

Before anything is moved, you can review the proposed changes. If you want to keep a specific folder structure, use category constraints such as a whitelist, or run AI File Sorter on a smaller folder where the intended structure is clearer.

For image support, AI File Sorter needs to connect to an LM Studio model that supports vision and exposes an OpenAI-compatible API. In LM Studio, make sure the vision model is loaded and that the local server/API is running.

In AI File Sorter, select LM Studio as the AI provider and choose a vision-capable model. If the app still asks you to download a model, it usually means AI File Sorter cannot see a compatible local model through the configured provider, even if one is loaded in LM Studio.

Check that:

• LM Studio's local server is enabled.
• The server URL in AI File Sorter matches LM Studio's API address.
• The loaded model supports image input, not text-only chat.
• The model name shown by LM Studio matches what AI File Sorter is configured to use.

Video support is different: AI File Sorter does not send entire videos directly to a vision model in the same way it sends images. Depending on the feature, video analysis may require a supported built-in workflow or may be limited to filename, metadata, or extracted-frame based processing.

AI File Sorter can use Ollama through the Ollama server, not by reading the model files directly from a folder such as B:\OllamaApp.

To use it:

1. Install and run Ollama.
2. Pull or create the model you want to use, for example with ollama pull llama3.1.
3. Make sure Ollama is running locally, usually at http://localhost:11434.
4. In AI File Sorter, choose the local/Ollama AI option and select or enter the model name.

The folder path, such as B:\OllamaApp, is only where Ollama stores its application or model data. AI File Sorter does not need that path; it communicates with Ollama through its local API.

Yes, partly. AI File Sorter lets you guide the structure by choosing how broad or detailed the categories should be, and you can use category constraints such as a whitelist when you want the results to follow a fixed set of folders.

Direct prompt files, such as a custom .md file with your own naming and folder rules, are not currently a built-in workflow. That said, this is a natural direction for future versions, especially for users who already have a preferred filing system and want AI File Sorter to follow it more closely.

Yes. If you sort a folder, then later add new files and run AI File Sorter on that folder again, the app can use its local categorization cache and learned review decisions to keep suggestions consistent. When the new files logically match categories and subfolders used before, AI File Sorter should suggest the same structure.

You will still be able to review the suggestions before applying them. For the most consistent results, keep the same settings, whitelist, and categorization mode between runs.

On macOS, the app may be blocked the first time because it is distributed outside the Mac App Store and is not signed with an Apple certificate.

Use the macOS help page linked from the download page for step-by-step instructions, including the Settings -> Privacy & Security approval step.

Full-disk sorting can take a long time because AI File Sorter has to scan many files, analyze their names and metadata, and sometimes run AI analysis on file contents.

Image analysis is usually the slowest part. Its speed depends heavily on your computer’s GPU. A discrete GPU with around 6 GB or more of video memory can make image scanning much faster. On systems without enough GPU video memory, or when the app falls back to CPU processing, image analysis can be much slower.

For large cleanups, it is often better to sort images separately from documents and other files.

Usually, no. AI File Sorter works best when used on specific folders or batches rather than an entire disk at once.

Good starting points are:

• Downloads
• Desktop cleanup folders
• Documents folders
• Picture folders
• External drive folders
• Project archives

For a first large cleanup, smaller batches are easier to review and usually more practical. Image-heavy folders are often best handled separately.

If the app is running a large analysis and your PC feels less responsive, the best practical options are:

• let the run continue when you are not actively using the PC
• stop or cancel the run if you need the computer back immediately
• restart later with a smaller folder or batch
• process images separately from non-image files

Windows will share CPU, GPU, and memory resources between apps, but local AI analysis can still be demanding, especially for images.

AI File Sorter does not currently include a dedicated low-resource or background-throttle mode.

No. AI File Sorter is not currently designed to run as a background service.

It is intended to be used as a batch tool:

1. Choose a folder.
2. Run the analysis.
3. Review the suggested categories and rename suggestions.
4. Approve the changes.
5. Close the app when finished.

This gives you control over what changes are applied to your files.

After the initial cleanup, most people should use AI File Sorter periodically rather than continuously.

A practical workflow is:

• run it on Downloads when that folder gets messy
• run it weekly or monthly depending on how many files you collect
• run it after importing a large batch of files
• run image-heavy folders separately if needed

The first large run is usually the most time-consuming. Later runs may be faster because the app can reuse its local cache for files it has already processed.

Yes. A newer version of AI File Sorter is planned with a faster visual AI model that should require less GPU video memory.

This should make image analysis faster or more practical for more users, especially on systems where the current visual model is too demanding.

Uninstalling AI File Sorter only removes the app. It does not delete the files or folders you sorted with it.

Windows:

1. Close AI File Sorter.
2. Open Settings > Apps > Installed apps.
3. Find AI File Sorter, choose the three-dot menu, then choose Uninstall.
4. If you used a portable download instead of an installer, delete the AI File Sorter folder or .exe file and remove any shortcuts you created.

macOS:

1. Close AI File Sorter.
2. Open Applications in Finder.
3. Drag AI File Sorter to the Trash, then empty the Trash if you want it removed immediately.
4. You can also delete the downloaded .dmg installer file if you still have it.

Linux:

1. Close AI File Sorter.
2. Delete the AppImage, extracted app folder, or launcher you installed.
3. If you installed it through a package manager, remove it with that same package manager.

If you also want to remove local settings or cache files, contact support and tell us which operating system you are using so we can give you the exact folder paths.