Peter Steinberger has gone from relative obscurity to being the toast of the AI ecosystem in three short months, but his path to joining forces with OpenAI had begun more than a decade ago.
When Peter Steinberger’s OpenClaw exploded onto the AI scene in late 2025 and ended up with an acquisition of sorts of by OpenAI, it appeared to be the work of a talented newcomer riding the AI wave. In reality, it was the latest chapter in a remarkable entrepreneurial journey that began nearly two decades earlier—and came after Steinberger had already built, scaled, and exited from one of Europe’s most successful bootstrapped technology companies.
The PSPDFKit Story: Building Without Funding
Long before OpenClaw, Steinberger had already proven himself as a builder. In 2010, while working as an iOS freelancer, he took on what seemed like a straightforward contract job: fixing a poorly coded PDF viewer that kept crashing. “The customer had already had it developed externally, but the program crashed every few minutes because it was simply badly programmed,” Steinberger recalled in a 2018 interview. “I decided to completely rewrite it and had the first iOS version after eight weeks.”
That contract work could have ended there. But when a colleague asked if he could use the engine too, Steinberger realized there might be a broader market. This was just after the release of the first iPad, when media companies were betting big on magazine apps built on PDF technology. “I set up a website with a shop within a day and announced it on Twitter. Already in the first week, I sold several licenses,” he said.
The timing became both blessing and curse. Steinberger had just accepted a coding job in Silicon Valley but was stuck waiting nine months for his visa. “I didn’t take on new freelance contracts during that time and therefore had time for my project,” he explained. By the time he finally started the San Francisco job, his side business was already earning more than his salary. Four months of night and weekend shifts later, he returned to Vienna to focus on what would become PSPDFKit full-time.
Scaling to Global Success
What followed was a masterclass in bootstrapped growth. Steinberger made the critical decision in 2013 to go all-in on becoming a multi-platform solution. “The strategic question was: Should the company stay small and try to establish itself sustainably in the niche, or should it become a multi-layer platform and focus on scaling?” he said. After market analysis, the answer was clear: “Either we become the number one solution for all platforms, or we disappear at some point.”
He brought on co-founders Jonathan D. Rhyne (COO) and Martin Schürrer (CTO) in 2014, expanded from iOS to Android and then web, and pivoted from a licensing model to subscriptions. The bet paid off spectacularly. By 2018, PSPDFKit had 40 employees, €5 million in annual revenue, and customers including Dropbox, Lufthansa, IBM, SAP, Atlassian, and DocuSign—all without raising a single euro in venture capital.
“We were actually profitable from day one,” Steinberger explained. “Over time, the offers kept increasing. We’ve sat at the table with major VCs. But if we can grow healthily from our own cash flow, why should we raise capital?” The company continued this trajectory until October 2021, when Insight Partners invested $116 million—the company’s first outside funding after a decade of growth. By then, nearly one billion people were using apps powered by PSPDFKit’s technology. Steinberger sold most of his shares in that transaction.
A Polyagentmorous Builder: Decades of Open Source
What many observers missed in the OpenClaw story was the foundation that made it possible: Steinberger’s decades of relentless experimentation with developer tools. His GitHub profile tells the story of a builder who never stopped tinkering, never stopped learning, and never stopped sharing his work with the community.
Long before OpenClaw became a household name in AI circles, Steinberger had created dozens of open-source projects spanning command-line tools, terminal enhancements, automation frameworks, and developer utilities. He describes himself as a “polyagentmorous builder”—someone who experiments with multiple agents and tools simultaneously, constantly pushing the boundaries of what’s possible.
His open-source portfolio reads like a master class in developer productivity. VibeTunnel transformed any browser into a terminal, letting users command agents from anywhere. CodexAsana provided intelligent agent limit monitoring. Peekaboo delivered lightning-fast macOS screenshots with GUI automation through MCP and CLI. His summarize tool could extract the essence of any URL or file instantly. RepoBar gave developers CI, PRs, and releases at a glance. The ag-cli tool brought Google services into the terminal—Gmail, Calendar, Drive, Contacts, Tasks, Sheets, Docs, Slides, everything accessible through commands. Poltergeist acted as a ghost that kept builds fresh through universal reload and file watching.
These weren’t just weekend projects thrown onto GitHub and abandoned. Each tool solved a real problem Steinberger faced in his own workflow. Each one demonstrated his philosophy of building tools that actually work, that integrate seamlessly with existing systems, and that respect the command line as a first-class interface.
This extensive portfolio reveals something crucial about the OpenClaw success story: it wasn’t his first viral hit or even his first experience building for developers. By the time he created OpenClaw, Steinberger had spent years perfecting the art of building developer tools that people actually want to use. He understood intimately what developers needed because he was solving his own problems first.
His legacy work also includes contributions that shaped entire ecosystems. CodeLooper was a macOS menubar app for Cursor workflow monitoring and automation. Interproject enabled modern Swift method swizzling with 10k+ stars on GitHub. The Aspects AOP library for Objective-C garnered similar acclaim. His PSPDFKit work, marked as “exited 2021,” represented the industry-leading PDF SDK that had already made him successful before OpenClaw was even conceived.
The Retirement That Wasn’t
After exiting PSPDFKit, Steinberger could have simply retired. Instead, he moved into what he called a “product research role” and began exploring new frontiers. In 2022, he co-founded Founders of Europe, an early-stage investment consortium with six other experienced startup founders, making pre-seed and seed investments.
But the real action was happening elsewhere. In May 2025, Steinberger founded Amantus Machina with a mission that would have seemed impossibly ambitious to most: “Working on the next generation of hyper-personal AI agents.”
This is where his decades of experience building developer tools, managing complex technical products, and understanding what businesses actually need converged with the new possibilities of large language models. Unlike many AI founders coming from pure research backgrounds or fresh out of university, Steinberger brought hard-won knowledge about what it takes to build products that people actually use at scale.
From Tinkering to Viral Sensation
The story of OpenClaw’s creation reflects Steinberger’s characteristic approach: start with a personal need, build something quickly, and iterate based on real usage. In November 2025, he wanted a way to check on his AI agents from his phone. “I just hacked together some WhatsApp integration that literally receives a message, calls Claude Code, and then returns what Claude Code returns,” he told Lex Fridman. “It took like one hour and it worked.”
What started as a personal tool evolved rapidly through real-world use. During a weekend trip to Marrakesh, he found himself using it constantly—not for coding, but for finding restaurants and getting local information. Then came the moment that crystallized the potential: he sent a voice message without thinking, even though he’d never built voice support. Ten seconds later, the agent replied normally.
When he asked how it had processed the audio, the agent explained it had identified the file format from the header, used FFmpeg to convert it, worked around a missing Whisper installation, found his OpenAI key in the environment variables, sent the audio via curl to OpenAI for transcription, and responded. “That was the moment where it clicked,” Steinberger said. “These things are damn smart, resourceful beasts if you actually give them the power.”
The Pattern of Success
Looking at Steinberger’s journey reveals a consistent pattern. With PSPDFKit, he identified a technical problem (reliable PDF rendering), built a solution for himself, recognized broader market potential, and scaled methodically without losing focus on product quality. With his dozens of open-source tools, he followed the same approach: identify friction in his workflow, build a solution, share it with the community, and iterate based on feedback. With OpenClaw, he applied this same proven playbook in a new domain: identify a need (managing AI agents conveniently), build a solution, discover unexpected capabilities through real usage, and let the product’s utility drive adoption.
Full Circle
When OpenAI announced Steinberger’s hire in early 2025, Sam Altman called him “a genius with a lot of amazing ideas about the future of very smart agents interacting with each other to do very useful things for people.” It was recognition of what Steinberger had already demonstrated through PSPDFKit, through dozens of open-source projects, and through OpenClaw: the ability to turn emerging technologies into products that millions of people use.
The fact that he chose OpenAI after serious courtship from both Meta and OpenAI—and after Mark Zuckerberg personally spent time tinkering with OpenClaw—speaks to both the value of what he’d built and his own understanding of where the technology could go. That OpenAI committed to keeping OpenClaw as an open-source project reflects Steinberger’s influence in the negotiations and his continued commitment to the community he’d built over decades of sharing his work.
For someone who bootstrapped a company from a freelance contract to a $116 million exit without venture capital, who built dozens of widely-used open-source developer tools, and who then created AI’s first viral agentic assistant a few years later, “retirement” was always going to be a relative term. What Steinberger brings to OpenAI isn’t just technical mastery—though he has that in abundance. It’s the rare combination of builder instincts, product sense, community engagement, and entrepreneurial experience that comes from spending twenty years turning ideas into products that work at global scale. The overnight success of OpenClaw was, like most overnight successes, decades in the making. It just took the right technology, the right moment, and the right builder—one who had spent years perfecting his craft through dozens of experiments, successes, and lessons learned—to make it happen.