Peter Steinberger on the importance of play in AI development

Peter Steinberger, the creator of the viral AI agent OpenClaw who has since joined OpenAI, argues that the most effective way to build with artificial intelligence today is through exploration and playfulness rather than rigid, pre-planned roadmaps. Reflecting on his own development process, Steinberger noted that he did not start with a master plan, explaining that much of his early work was simply exploration. He advises builders to experiment without expecting to become immediate experts, emphasizing that the best approach to building with AI today is to remain playful and open to learning. He suggests that developers should focus on building things they have always wanted to create, allowing themselves the time to improve.

Steinberger discussed his journey on the first episode of Builders Unscripted, a new podcast by OpenAI, where he spoke with Romain Huet, the company’s Head of Developer Experience. He explained that he initially built a tool integrated with WhatsApp but paused development, expecting major AI laboratories to release a similar product. However, by last November, he was surprised to find that no AI labs had started to build what he wanted, prompting him to create the initial prototype for OpenClaw. He realized through this process how capable modern AI models have become at solving problems autonomously.

During the podcast, Steinberger strongly rejected the popular industry term vibe coding, which describes using AI to write software. “I think vibe coding is a slur,” said Peter Steinberger, the creator of the viral AI agent OpenClaw, suggesting that the term minimizes the actual skill required to work effectively with AI models. He compared the process of learning to code with AI to learning to play an instrument, noting that a person is not going to be good at the guitar on their first day. Instead of giving up when initial prompts fail, developers must adapt and build a feel for how the models respond.

As traditional software development methods change, Steinberger believes that individuals who are high agency—meaning they are proactive and take responsibility for outcomes—will remain highly sought after. He argued that smart, proactive individuals who focus on solving problems and creating new things will experience higher demand than ever before, even as AI tools become more capable.

Why it matters

As artificial intelligence tools become more accessible, the shift from traditional coding to agentic workflows requires a fundamental mindset change. Steinberger’s experience highlights that adaptability and proactive problem-solving are becoming far more valuable than rigid technical planning.