Claude Cowork resize images, convert them to WebP and rename them

By Stefan Petri Last updated on November 14, 2026

Imagine you have a folder of photos, each several megabytes in size, and you don't feel like opening every single one in Photoshop. This is exactly where Claude Cowork comes in. The feature works directly on your hard drive, shrinks images, converts them into modern formats and renames them based on their content. In this guide, I walk you step by step through the entire process, from the backup to checking the finished files.

Play video from YouTube

The video is only loaded from YouTube when you play it – data (including your IP address) is then transmitted to YouTube/Google. More in the privacy policy

Introduction

Many images for websites, newsletters or social media have a resolution and file size that are far too high. As a result, a website loads slowly, and file names like "IMG_4721.jpg" help neither you nor search engines. Claude Cowork takes over this routine work while you can take care of more important things. It is important that you go through the steps in the right order. Especially the backup before you start will save you a lot of trouble later.

Key Takeaways

Key Takeaways

• Claude Cowork works directly on your file system, so you should create a backup before every task.
• You give the task in plain language, for example "Scale all images down by 25%".
• Converting to WebP significantly reduces file size while maintaining the same image quality. AVIF is often even smaller.
• The AI can rename images based on their subject, which makes the file names SEO-friendly.
• After renaming, you can check the files directly and have unnecessary intermediate folders deleted again.

Step-by-Step Guide

Step 1: Create a backup before you start

Before you enter even a single command, back up your original images. Claude Cowork has direct access to your files, and even the best AI can make mistakes. It is best to create the backup folder outside the working folder, for example one level above it or in a cloud such as Dropbox. In the example in the video, the AI had deleted too much when dealing with duplicate images and had to be restored from the backup. A copy costs you only a few seconds, whereas a lost collection can cost you hours.

Step 2: Select the working folder

In the next step, you specify which folder Claude Cowork should work with. To do this, open your file manager, select the folder with the images and copy it with Ctrl+C. Then you paste it into Claude Cowork via the selection dialog and confirm the selection. Make sure that you really pick the folder with the test images or your copies, not the original collection.

Step 3: Allow access

As soon as the folder is selected, Claude Cowork asks for permission. You can either allow access just once or grant it permanently. For a one-off task, one-time permission is completely sufficient. If you use the folder regularly for similar tasks, you can choose permanent access. So decide consciously, because the permission determines what the AI is allowed to do with this folder in the future.

Step 4: Formulate the first task

Now you enter the command. Phrase it as clearly as possible. In the example, the task reads roughly: All images should be limited to a width of 800 pixels where appropriate, and uniformly scaled down by 25%. Start with a single work step first so that you can check the result before you assign more complex tasks. Confirm the input with Enter.

Step 5: Choose the right model

At the bottom of the input area, you select the AI model. The strongest models often deliver better results but also consume more tokens. For simple tasks such as shrinking, a smaller model is usually enough. For more complex tasks such as cropping subjects, the more powerful model is worthwhile. Note that you can no longer switch the model during an ongoing session. So decide beforehand which model suits your task.

Step 6: Follow the progress

While Claude Cowork is working, you can use the activity display to see exactly what is happening in the background. There you can see, for example, that the AI writes a Python script and processes the images one after another. In the example, 26 images had to be processed. The AI then reported that it was finished and offered to overwrite the originals directly or create a new folder. You should always accept the latter offer, because that way your originals are preserved.

Step 7: Check the result in the new folder

Claude Cowork has created a new folder containing the resized images. Open this folder and compare the resolution and file size with the originals. In the example, the largest image was only around 11 MB after processing, and the resolution was about 3000 instead of 4000 pixels. Open individual images in the preview window to check whether the quality is right.

Step 8: Convert to WebP format

In the next task, you ask Claude Cowork to convert the resized images to the WebP format. WebP was developed specifically for the web and can be stored significantly more space-efficiently than JPEG or PNG. In the example, the folder size dropped from around 60 MB to 17 MB, and the largest image, which previously weighed 10 MB, was afterwards well under 1 MB. At the same resolution, the image looked exactly the same. Here too, the AI creates a new folder so that you can compare the results.

Step 9: Comparing file sizes

Now compare the folder sizes of the different versions. To do so, open the properties of the respective folder in the file manager. This way, you can immediately see how strong the effect of the conversion is. Especially with larger image collections, the difference quickly adds up to a noticeable gain in storage space and loading time.

Step 10: Testing the AVIF format

There is another modern format called AVIF, which is often even smaller than WebP. In the video, it was tried out as a test, and the AI converted the collection without any problems into yet another new folder. Current browsers now support AVIF and WebP well. However, older browsers may have problems. For websites, a so-called fallback is therefore recommended: the browser is offered the most modern format, and if it cannot display it, it falls back on a standard format such as JPEG. For this, however, you then have to keep the image available in several formats.

Step 11: Stay patient with larger batches

With extensive jobs, processing can stall. In the example, the AI got stuck at six out of 26 images. After a waiting period, however, everything was finished. Afterwards, reload the folder to make sure that all files have arrived. In the practical example with around 300 drone shots, the AI had to be prompted again on the first attempt because the cropping of the solar modules did not quite succeed. A more powerful model and a more precise instruction solved the problem.

Step 12: Have images renamed automatically

Now comes the part that saves especially much time. Ask Claude Cowork to rename the images based on their content. The AI looks at each image and assigns a descriptive file name, for example "Woman-with-hat-at-the-lake" or "Desert-sand-dune-sunset". Such names are not only clearer, but also significantly more helpful for search engines than cryptic camera names. While the AI is working, you can, by the way, already start new tasks; both processes run in parallel.

Step 13: Check the results

After renaming, check a few examples. To do so, hover the mouse over the files to see a preview. In the video, most names were right on the first try, but with one image the subject still had to be rotated. Especially with images with many details, a spot check is therefore worthwhile. The AI can additionally create an overview for you of which file it named how.

Step 14: Clean up unnecessary files

During processing, auxiliary files such as thumbnails or intermediate folders often arise. Claude Cowork may ask you whether it should delete such files. Check the list carefully before you confirm, and make sure that no originals are among them. In the example, the thumbnail folder was automatically removed as well after confirmation. Only then is the working folder clean again.

Step 15: Try out further use cases

Claude Cowork can do significantly more. You can bring images into specific aspect ratios, such as 9:16 for Instagram Stories or Reels, or 16:9 for YouTube. Cutting out subjects and combining several images are also possible, although the results turn out differently depending on the task. Try out such tasks at your leisure to get to know the limits of the function.

Summary ??? Resizing, converting, and renaming images with Claude Cowork

With Claude Cowork, you can reduce the size of image collections in just a few minutes, convert them into modern web formats such as WebP or AVIF, and rename them sensibly based on the subject. It is important that you create a backup beforehand, deliberately select the working folder, and then spot-check the results. Anyone who follows these steps saves an enormous amount of time compared to manually editing individual images.

Trainer About Stefan Petri Together with his brother Matthias, Stefan Petri runs the popular specialist forum PSD-Tutorials.de and the e-learning platform TutKit.com, which focuses on the training and further education of digital professional skills. All courses by Stefan Petri →