Peter Steinberger, the creator of the viral AI agent OpenClaw who has since been hired by OpenAI, has some advice for those experimenting with AI technology, including AI agents.

From his own experience, the best way to build today is to explore, be playful, and not expect to be an expert at what you do right away.

“I wish I could say that I had the unified plan in the beginning, but a lot of it was just exploration,” Steinberger said.

“I wanted things, and those things didn’t exist, and … let’s say I prompted them into existence.

” The developer was chatting with OpenAI’s Head of Developer Experience, Romain Huet, on the first episode of the company’s new Builders Unscripted podcast.

Here, he spoke about what OpenClaw was like in its early days and how he didn’t have a plan when he got started. Steinberger explained he began by building a tool that would integrate with WhatsApp, but then set it aside for a bit and focused on other things, as he assumed the AI labs would build something like what he was working on in the near future.

“I just experimented a lot.

My mission was, kind of like, to have fun and inspire people,” Steinberger noted. By last November, however, the developer was surprised that no AI labs had started to build what he wanted to use. That led him to create the initial prototype of what’s now OpenClaw.

“Where it really clicked was where I was at this weekend trip in Marrakesh, and I found myself using it way more because it was so convenient … There was no really good internet.

[But] WhatsApp just works everywhere,” he said. The tool made it easy for him to find restaurants, look up things on his computer, send texts to friends, and more. The more he played with the technology, Steinberger realized how good modern AI models have become at problem-solving, much like coders are. “Now they can just, like, actually come up with the solutions themselves, even though you never programmed them at all,” he noted. Throughout the process of building, Steinberger said that his workflow improved — and he stresses to other developers that’s something that can take time, so don’t give up.

“There’s these people that … write software in the old way, and the old way is going to go away,” he pointed out.

They then decide to try vibe coding but are disappointed with the results.

“I think vibe coding is a slur,” said Steinberger, basically suggesting that it’s not as simple a process at first as the term makes it sound.

“They try AI, but they don’t understand that it’s a skill,” he said, then compared the process of coding with AI to learning guitar.

“You’re not going to be good at guitar on the first day,” he said. Instead, he recommends that people approach learning with a more playful attitude.

If he writes a prompt now, he has a gut feeling as to how long it will take, and if it takes longer, he reflects on what may have gone wrong and adapts.

“My … advice always is, approach it in a playful way.

Build something that you always wanted to build. If you’re at least a little bit of a builder, there has to be something on the back of your mind that you want to build.

” This ability to experiment and have fun is what’s most important, especially at a time when people are worried their jobs will be overtaken by AI.

“If your identity is: I want to create things.

I want to solve problems. If you’re high agency, if you’re smart, you will be in more demand than ever,” Steinberger said

Highlighted sentences link to their corresponding claims. Click any highlighted sentence to jump to its detailed analysis.
Highlight Colors Indicate Claim Quality:
✓ Healthy Claim - No fallacies or contradictions detected
⚠️ Minor Issues - Has contradictions or minor fallacies
🚨 Serious Issues - Multiple contradictions or severe fallacies
Quality Criteria: Claims are evaluated for logical fallacies and contradictions with other news sources. Green highlights indicate healthy claims suitable for reference.
Source