If you use OpenClaw, you are running software that was prototyped in roughly one hour by a burned-out Austrian developer who had just sold his company for a reported €100 million and booked a one-way ticket to Madrid to figure out what to do with the rest of his life. That developer is Peter Steinberger — known online as steipete — and his path to building OpenClaw is one of the more interesting origin stories in recent open source history.
A Computer in Rural Austria
Steinberger was born in 1986 and grew up in rural Austria. When he was 14, a summer guest brought a personal computer into his world. That was enough. He became obsessed with software, studied at Vienna University of Technology, and eventually landed a senior iOS engineering role in Silicon Valley — while also teaching mobile development back at his alma mater. He split his time between London and Vienna, and by his own account, spent most of his twenties and early thirties coding.
PSPDFKit: The Billion-Device Business
In 2011, while waiting six months for a U.S. work visa, Steinberger started bootstrapping a side project: a PDF rendering framework for iPads. The problem was, as he described it, simple yet incredibly difficult. That project became PSPDFKit — a document processing SDK that quietly became the industry standard, powering PDF functionality for Apple, Dropbox, and hundreds of other companies. By the time he exited in 2023, PSPDFKit was running on over a billion devices.
The exit was worth a reported €100 million. It was also, by his account, a relief. The years spent as CEO had ground him down — board meetings, co-founder tensions, the relentless pace of customer demands. He later described the feeling: “I felt like Austin Powers where they suck the mojo out. I couldn’t get code out anymore. I was just, like, staring and feeling empty.”
Madrid, Therapy, and an Empty Feeling That Wouldn’t Leave
After the exit, Steinberger tried the standard recovery playbook: travel, therapy, distance from screens. He booked a one-way ticket to Madrid. None of it resolved the emptiness. For someone who had been writing code since adolescence, the loss of the ability to build things was not a productivity problem — it felt like a loss of identity.
The turning point came in April 2025, while he was attempting to build a Twitter analysis tool. He ran into AI-assisted coding in a serious way for the first time and recognized it immediately as a paradigm shift. Large language models could handle the repetitive, draining parts of coding — boilerplate, debugging loops, context-switching overhead — and leave the higher-level problem-solving intact. The mojo came back.
Project 44: Prompted Into Existence in One Hour
OpenClaw was Steinberger’s 44th AI-related project since 2009. It started, like many of his projects, with annoyance. He wanted a locally-running AI agent that could manage emails, control a browser, and be reached through messaging apps like WhatsApp and Telegram. Nothing that fit his vision existed. “I was annoyed that it didn’t exist,” he told Lex Fridman, “so I just prompted it into existence.”
The prototype took about an hour. He published it on GitHub in November 2025 under the name Clawdbot — a playful nod to Anthropic’s Claude, which he had been using extensively. The project was built around a local-first philosophy: the agent runs on the user’s own hardware, memories are stored in plain Markdown files, and nothing is pushed to a corporate cloud unless you choose it.
The Rebranding Saga
Clawdbot grew fast — faster than Steinberger had planned for. And with visibility came a trademark complaint from Anthropic, who objected to the Clawd name in proximity to their own Claude branding. On January 27, 2026, the project became Moltbot — keeping the lobster theme, since molting is what lobsters do when they grow. Three days later it became OpenClaw, because Moltbot, as Steinberger acknowledged, never quite rolled off the tongue.
By the time the dust settled on the naming, the project had over 145,000 GitHub stars, 2 million visitors in a single week, and a place in the conversation about what open-source AI agents could actually look like at scale.
Altman, Zuckerberg, and the Decision
The growth attracted attention at the highest levels. Sam Altman called him a “genius with a lot of amazing ideas.” Mark Zuckerberg made personal outreach — what sources described as likely a six-figure-plus offer to bring the project under Meta’s umbrella. Steinberger chose OpenAI, citing the need for access to “the latest toys” to scale his vision. He was clear that the decision was not primarily financial: “I don’t do this for the money. I want to have fun and have impact.”
To address concerns about an open-source project being absorbed by a centralized company, Steinberger announced that OpenClaw would transition to an independent open-source foundation with OpenAI as a supporting partner rather than an owner.
What He Built, and Why It Matters
OpenClaw is not Steinberger’s most technically complex project — PSPDFKit, running on a billion devices across a decade, was almost certainly harder to build and maintain. What makes OpenClaw significant is the timing and the philosophy. It arrived at the moment when local AI agents became practically viable, built around a local-first architecture that was a deliberate counterpoint to the cloud-first default of most AI tooling. The SOUL.md and AGENTS.md approach — storing agent identity and memory in plain Markdown files — is idiosyncratic and, to many users, immediately legible in a way that proprietary agent formats are not.
There is also the personal thread. Steinberger described being moved by watching agents solve problems he had not explicitly programmed for — transcribing a voice message without being asked, checking on his health after surgery. For someone who spent years unable to feel anything when looking at code, that was not a small thing.
He now describes himself not as a CEO or a founder, but as a “full-time open-sourcerer of the agentic revolution.” It is a characteristically odd turn of phrase, and probably exactly right.
Leave a Reply