How OpenClaw Went From Personal Project to 4.7M Users

Key Insights

• OpenClaw started as a personal frustration: needing a way to send prompts from a phone to monitor agents on a computer
• The project went viral in January after a single Discord demo, reaching 4.7 million weekly downloads at its peak
• Success brought unexpected consequences—security criticism, founder burnout, and the challenge of managing 9,500+ configuration options
• The real competitive advantage wasn't the feature set, but the fun of building it and the open-source ethos behind it
• Focus and personal brand matter more than perfect code or venture funding

From Annoyance to Inspiration

Peter Steinberger built OpenClaw to solve a personal problem. In November 2024, frustrated by the lack of a simple way to relay prompts from his phone to monitor agents running on his computer, he spun up a solution. In an hour, his coding agent built him a WhatsApp relay—something that felt "magical" not because it was complex, but because it simplified the underlying complexity away. You didn't need to think about which model to use or when to start a new session. The experience felt less like a tool and more like a friend.

What made OpenClaw different wasn't the initial functionality. Developers had been getting similar results for months by typing into terminals and getting AI replies. The magic was in the feeling—the simplicity and the emotional connection. But when Steinberger tried to explain this to his Twitter audience, people didn't get it. So he did what annoyed people do: he created group chats with friends and let them talk to his WhatsApp Relay. The reaction was strong. Some people were amazed; others were scared. That emotional response was the real signal—and a clear indicator of product-market fit.

The Viral Moment

A pull request to add Discord support transformed everything. Steinberger initially questioned the request (it was called "WhatsApp Relay," after all), but eventually decided "what the hell?" and rolled it out. Around New Year's Eve, he created a Discord server and put his agent in it. He called it Clawd and stayed up all night letting people interact with it, watching them try to hack it, seeing them finally get it.

He had built the software to be resilient—it was a launch daemon. When he pressed Control-C to shut down at 7 AM, the agent restarted automatically. He slept for 10 hours, expecting peace. Instead, his inbox flooded with 800 messages. People had tried to hack it. His heart sank. But as he read through, nothing had actually broken. This was the moment it went viral.

Within weeks, reporters called in the middle of the night. His email became a waterfall. Mac Minis sold out (and still haven't come back). He received more podcast invites in one month than in his entire 39 years before. The project went through name changes—Clawd became Clawbot, then OpenClaw. Within eight months, more than 18,000 different people opened issues or pull requests, with almost 3,000 contributors. The numbers didn't feel real. Jensen Huang called it "the most successful open-source project in the history of humanity."

But viral success nearly broke him. Reporters, security researchers, and curious humans flooded his inbox. Someone leaked his phone number and private details. He stopped answering friends. He didn't want to look at his phone anymore. For a moment, he was close to deleting everything.

The Cost of Scaling

With success came unexpected pressure. The project faced waves of security criticism. A press report claimed 20% of skills were malicious—but Steinberger's actual research showed 0.3%. He scanned all 67,000 skills and shipped a paper, but "a correction never travels as far as a scare." He felt responsible. He hardened the codebase, added sandboxing, allow lists, web protocols with permissions. Most users didn't care about these details, but the work was real and slowed development.

The feature explosion was brutal. By the end, OpenClaw had around 9,500 configuration options when counting all permutations. You can't test all of them. Every new feature meant a new configuration option—no one wanted their setup broken. Meanwhile, other companies with VC funding pushed harder. When Anthropic disabled their subscription service with 24 hours notice, OpenClaw, which had optimized for Opus, got hit hard. Users moved on.

By May, downloads bottomed out at 835,000 per week. Then, after being "declared dead" in June, the project peaked at 4.7 million weekly downloads.

What Went Wrong

Somewhere around February, building stopped being fun. It became a responsibility. Steinberger found himself with two jobs: open source and a calling. Worst of all, he stopped using his own product. When you don't dogfood what you ship every day, "it gets shit." He shifted from building something he loved to building something for everyone.

His split attention meant he became the person who fixed bugs, handled security issues, and provided support. He was so deep in everything that he didn't think strategically. He could have asked for help or moved responsibility, but he didn't.

What really hurt was the dependency problem. Steinberger had optimized OpenClaw for specific models, particularly Opus. When Anthropic's business model changed, there wasn't time to adapt. The lesson: "Your dependencies' business model is your business model."

The Turnaround

Around May, things shifted. Steinberger found amazing people who believed in open source. Nvidia stepped in early and asked what he needed, then sent people to take over security work. His visa got sorted. The nonprofit took shape with amazing donors. The joy of building started coming back.

The press declared an "OpenClaw killer" nearly every week—he counted 20 of them. But none of the killer stories understood what OpenClaw was really about: open source, solving things that annoyed him, and the fun of building. "It's hard to compete with someone who's just there having fun. Fun is velocity."

What's Next

OpenClaw now has ten people on payroll and is hiring, including for a CEO role. Steinberger is proud that the project moved AI from something "nebulous and scary" into something "fun and weird." There are 33,000 claw-named repositories across the internet—the idea has taken on a life of its own.

The original vision remains: an always-on agent that doesn't require constant attention. The landscape is evolving faster than the software, but that's opportunity. Steinberger is building OpenClaw with OpenClaw itself—where agents see each other's sessions on a team server and can orchestrate work together. The dream is to finally move beyond people running terminals with their laptops open because the agent needs to keep working.

The Three Things to Remember

1. Don't stop having fun. Fun is your ultimate driver for getting the best ideas.
2. Listen to your gut and fix the things that annoy you. It might become the next big thing.
3. Stay focused. Another podcast won't make you win. Live in the future, build what's missing, and keep shipping—it confuses them.

Conclusion

OpenClaw's rise from a personal WhatsApp relay to a 4.7-million-user phenomenon wasn't about perfect planning or venture funding. It was about one person solving an annoying problem, staying true to that vision, and letting others contribute to something they genuinely enjoyed building. The real lesson isn't about viral growth—it's that your personal brand and the fun you have while building matter far more than any single feature or product. When you stop having fun, users feel it. When you're having a blast solving problems, they feel that too.

---

Original source: Peter Steinberger: What Happens When 4.7 Million People Let It Cook

powered by osmu.app