LLM Designs n8n Workflows Best?

In a recent YouTube video by Farzad (AI RoundTable), four major Large Language Models (LLMs) – Claude, GPT, Gemini, and Mistral – were tested to evaluate their effectiveness in designing workflows using n8n. The task involved building an automation workflow that monitors YouTube channels, extracts and summarizes transcripts, and sends the results to Discord. As the video unfolds, the differences in performance between the models become apparent. Claude and Gemini emerged as the strongest contenders, providing a solid foundation for building n8n workflows, though manual intervention was still necessary for optimal results. Conversely, GPT and Mistral struggled, failing to deliver functional workflows from start to finish. The process underscored the importance of updating LLMs with current documentation for best results – a factor critical to improving model output. Farzad emphasizes that supplying updated and relevant documentation significantly enhances workflow outcomes, transforming LLM-generated drafts into near-production-ready workflows. The presenter illustrates this through a live demonstration, showcasing the workflow in action and highlighting practical benefits from the models’ initial inputs.