AI Creator uses Music Video Using Code Instead of Video Tools

An AI creator has demonstrated a different approach to producing music videos, using artificial intelligence to write the code behind the visuals rather than generating conventional video clips.

The creator, known online as klöss, shared the experiment on 19 October 2026 after asking Claude Opus 5.5 to produce a music video using HTML, GSAP and WebGL.

The project began with an AI-generated music track combining Mongolian throat singing and dubstep, created using Suno v6.

Claude then spent nine hours and 34 minutes working through the development process before delivering the finished visuals.

How was the AI music video created?

Instead of using a text-to-video generator, klöss instructed Claude to build the production through web-based programming tools.

The creator supplied the musical track and asked the AI system to develop a video entirely through code. HTML was used to provide the underlying structure, while GSAP, a JavaScript animation library, helped control the movement and timing of visual elements.

WebGL provided technology for rendering graphics, allowing imagery to be created and displayed through a browser-based environment.

The approach differs from AI video generators that produce finished clips directly from written descriptions. In this case, the AI was responsible for developing code that could produce the visuals, rather than relying entirely on generated video footage.

Claude worked for more than nine hours

According to klöss, the AI system spent nine hours and 34 minutes developing the project. During that time, Claude repeatedly reviewed its work and made adjustments as part of an extended coding process.

The creator described the system as auditing itself and working through repeated cycles before producing the finished result. Following the experiment, klöss shared the video and the prompt that Claude had developed, encouraging other creators to experiment with the technique.

The reported duration relates to the AI's development process and should not be interpreted as the total amount of human work required for a conventional music video.

Why is this different from AI video generation?

Most AI filmmaking tools discussed in entertainment news create images or moving footage directly from text instructions.

Code-based visual production works differently because the resulting imagery is generated through programming instructions.

That approach can give creators control over elements such as movement, positioning, timing and visual effects, depending on how the code has been designed.

It can also allow animations to be adjusted by changing particular instructions rather than regenerating an entire video clip. However, this does not automatically make code-generated visuals better or easier to produce than conventional animation or generative video.

The practical value depends on the type of imagery being created, the technical skills available and the creative goals of the production.

Could this approach help independent filmmakers?

The experiment is particularly relevant to creators interested in combining music, animation and digital effects without following a conventional production workflow.

By instructing an AI assistant to write the code, someone with a strong visual idea may be able to experiment with techniques that would otherwise require substantial programming experience.

However, evaluating the finished work and making creative decisions remain important responsibilities. Klöss's demonstration also does not establish that the result was an interactive experience controlled by viewers. The reported achievement concerns generating the music video's visuals through code.

The project provides another example of how AI-assisted filmmaking is developing beyond conventional text-to-video tools, with creators experimenting with programming systems as a way of producing original entertainment.

Where can you follow klöss?

Those interested in exploring more of klöss's AI experiments can follow the creator on X (@kloss_xyz), where they share projects involving artificial intelligence, coding and digital creativity.

The full demonstration of the music video, including details of the nine-hour development process, is also available through What People Made, alongside another experiment exploring AI-generated animation.