Shopify CEO uses AI tool Claude to read his MRI scans

MRI scans are stored in DICOM format and need special software to view. (Image credit: File image)

• Shopify CEO used AI to build a web viewer for his MRI scans
• Claude AI generated code for viewing DICOM files in a browser
• AI tools help make medical imaging more accessible for patients

Did our AI summary help?

Shopify chief executive Tobi Lütke recently shared how he used an artificial intelligence tool to read and view his own MRI scan files, bypassing the specialised software normally required for the task.

Lütke explained in a post on X that he had received his annual MRI scans but did not want to install the Windows-based software that usually comes with medical imaging files. Instead, he turned to Claude, the AI assistant developed by Anthropic, to help build a quick alternative.

MRI scans are typically stored in a format known as DICOM, which requires dedicated medical imaging software to view properly. These programs are often designed for doctors and hospitals, and they can be complicated for ordinary users to install and navigate.

Rather than using the supplied software, Lütke asked Claude to help create a small web-based viewer that could open the files directly in a browser.

Using AI-generated code, he built a simple HTML tool that allowed him to load the MRI data and scroll through the images more easily. According to Lütke, the result was cleaner and easier to use than the software that came with the scans.

He also said the tool was able to add annotations to the images based on the scan’s information, helping highlight certain details within the medical images.

The episode highlights how AI coding assistants are increasingly being used for small, practical tasks by people with technical backgrounds. Instead of writing software line by line, developers can describe what they want and allow the AI system to generate the basic code.

Tools such as Claude, GitHub Copilot and other AI programming assistants have become common among engineers for tasks like building prototypes, debugging code and creating simple applications.

Lütke’s experiment does not mean AI should replace medical professionals when interpreting scans. MRI images are complex and normally require trained radiologists to analyse them properly.

But the example shows how AI tools are beginning to make specialised digital formats easier for individuals to work with.

Medical imaging files have long been difficult for patients to access outside hospital systems. If tools like this become more common, patients may eventually

find it easier to view their own scans and medical records without relying entirely on proprietary software.

For now, Lütke’s quick AI-built viewer is more of a demonstration of what modern coding assistants can do than a replacement for professional medical tools.