Ich hab mich an KI angepasst - und bin trotzdem überflüssig

Summary

The video demonstrates how the creator built a prototype “design machine” in a few days. It also raises a concern: adapting to AI may not protect designers’ jobs if these tools can automate much of their work.

How the design machine works

• AI works directly in Figma: The creator connects a Claude-based AI agent to Figma. Instead of building layouts manually, she gives it plain-language instructions, sometimes by voice, and it creates or adjusts editable design elements.
• Natural-language instructions are enough: In a live demo, she asks the agent to create a promotional “hero” block and later to center its text. She also tries vague requests such as “make the buttons more fancy.” The agent often infers what she means, suggests the relevant design term, and makes the change.
• Branding and website creation: For fictional businesses, AI helps generate a name, logo, corporate identity, color palette, typography, and website designs. The creator says it also added features, such as animations, without being asked.
• Automated campaign assets: The prototype analyzes a website, extracting details such as its logo, colors, tone, fonts, imagery, and popular products, then suggests campaign concepts. The user can choose formats such as newsletters, Instagram posts, or posters, and opt to use AI-generated images.
• Multiple versions and self-checking: The machine generates three variations for each selection, builds the designs in Figma, then runs a review-and-correction pass. The creator says she added design rules to catch image errors—such as extra fingers or malformed animals—and regenerate images when needed.
• Pinterest as a design reference: The system searches Pinterest for relevant inspiration to guide its layouts, rather than relying only on rules supplied by the creator.

Demos, results, and limitations

The creator tests the system on two fictional brands: an esoteric-satire project and a bakery. It produces newsletters and social-media designs for the first, then newsletters and posters for the bakery. She finds the results usable and sometimes surprisingly polished, though she points out details she would personally refine, such as colors, image sizes, and recurring layout motifs.

She emphasizes that this is a high-level prototype, built quickly rather than as a finished commercial product. Some formats have had little training or refinement, and the process is not fully automated: the Figma output is editable, but exporting and other newsletter-production steps remain separate. She believes additional design rules and iteration could improve the results.

Main analysis

The creator’s central concern is that AI does not need to match a designer’s standards perfectly to replace much of the work. In her experience, many clients and end customers accept designs that look “good enough,” particularly when the message or offer is clear. Once a company’s design rules are defined, she argues, one person—or a project manager—may be able to generate work that previously required a larger design team.

She describes having few tasks at her company and says that learning to use AI may amount to teaching it how to take over her job. The video challenges the advice to “adapt” by asking what workers gain if adapting leads to an automated process that reduces the need for them. She also questions where displaced designers, copywriters, illustrators, programmers, and others are expected to go.

Video format and sources

The video is a demonstration and commentary rather than a formal product review or step-by-step tutorial. It includes live examples of AI-assisted design workflows and a discussion of their strengths, limitations, and employment implications.

The creator, a designer who works in newsletter production, narrates the demonstrations and analysis. The tools shown or discussed include Claude, ChatGPT, Figma, and Pinterest.