OpenClaw: the open-source AI assistant that started as a weekend hack

There's a lobster running on a Mac Mini somewhere, checking someone's email, managing their calendar, and writing code on their behalf. No, this isn't a fever dream. It's OpenClaw — and it might be the most important open-source project to emerge from the AI agent era so far.

From WhatsApp relay to 100,000 stars

The story begins in late 2025, when Peter Steinberger — known for decades of impactful open-source contributions in the iOS and developer tools space — hacked together a weekend project. The idea was simple: relay messages between WhatsApp and an AI model so he could talk to Claude from his phone.

That script worked. Then it kept working. Then other people wanted it.

What started as a personal convenience became something much larger. Within weeks, the project had a Discord server full of contributors, a growing plugin ecosystem, and a trajectory that no one — including Steinberger — had predicted.

By late January 2026, the project had crossed 100,000 GitHub stars and drawn over 2 million visitors in a single week. For context, most successful open-source projects take years to reach that kind of adoption. This one did it in roughly two months.

The naming journey

The project's naming history is worth telling because it mirrors its evolution.

It was first called Clawd — a playful portmanteau of "Claude" and "claw." It felt right until Anthropic's legal team politely asked for a change. Fair enough.

Next came Moltbot, chosen during what Steinberger describes as a chaotic 5 AM Discord brainstorm with the community. Molting — the process by which lobsters shed their shells to grow — was a meaningful metaphor. But the name never quite stuck.

On January 29, 2026, the project landed on its final identity: OpenClaw. The name captures both pillars of what the project has become. Open — open source, open to everyone, community-driven. Claw — the lobster heritage, a nod to where it all started.

This time they did the homework: trademark searches, domain registrations, migration code. The lobster was here to stay.

What OpenClaw actually does

At its core, OpenClaw is an open agent platform that runs on your machine and connects to the chat apps you already use — WhatsApp, Telegram, Discord, Slack, Signal, even iMessage.

But calling it a "chatbot" undersells it dramatically. OpenClaw can:

• Browse the web — fill forms, extract data, navigate sites autonomously
• Access your file system — read, write, and execute scripts (with your permission, sandboxed or full-access)
• Remember you — persistent memory that makes the assistant uniquely yours over time
• Run on any model — Anthropic, OpenAI, or fully local models; your keys, your choice
• Extend itself — a plugin and skills system where the community builds capabilities, and the assistant can even write its own

The philosophy is one sentence: your assistant, your machine, your rules. Unlike SaaS AI products where your data lives on someone else's infrastructure, OpenClaw runs where you choose — laptop, homelab, VPS, Raspberry Pi. Your conversations, your memory, your data stay yours.

Why the creator matters

Peter Steinberger is not a newcomer to open source. He built PSPDFKit, contributed to the iOS and macOS ecosystem for over a decade, and has a track record of shipping software that other developers trust. That credibility matters here.

Building a personal AI assistant that has access to your email, calendar, files, and browser requires trust. Trust that the code does what it says. Trust that there are no hidden data pipelines. Trust that the person behind the project cares about getting it right.

Steinberger brought that trust with him. But more importantly, he did something unusual for a solo creator with momentum: he actively worked to distribute ownership. Within days of the project's explosive growth, he began onboarding maintainers, establishing processes for the flood of pull requests and issues, and figuring out how to pay contributors — full-time if possible.

This is not a one-person project anymore. It's a genuine community effort, and the speed at which that community self-organized says something about the quality of both the software and its leadership.

The community that built itself

Open-source projects often struggle with the gap between users and contributors. OpenClaw collapsed that gap almost immediately.

People didn't just install it and use it. They built on it. Custom skills for Todoist integration, WHOOP health data, Spotify control, flight searches, home automation. One user had their OpenClaw instance build a website from a phone while putting a baby to sleep. Another had it autonomously monitor Sentry for errors, reproduce them, and open pull requests.

The Discord server became a living workshop where people shared skills, debugged setups, and pushed the boundaries of what a personal AI assistant could do. The ethos was hackable by design — your context and skills live on your machine, not in a walled garden.

As one community member put it: "It will actually be the thing that nukes a ton of startups, not ChatGPT as people meme about. The fact that it's hackable — and more importantly, self-hackable — and hostable on-prem will make sure tech like this dominates conventional SaaS."

Why open source is the only way this works

There's a deeper principle at play. Personal AI assistants, by definition, handle your most sensitive information — emails, health data, financial documents, private conversations. The only architecture that respects that reality is one where you can read every line of code, run it on hardware you control, and verify that nothing leaves your machine without your explicit consent.

Closed-source personal assistants ask you to trust a company. Open-source ones ask you to trust the code. In a world where AI is becoming the interface for everything — your work, your home, your health — that distinction matters more than ever.

OpenClaw isn't the first open-source AI project, and it won't be the last. But it may be the first to demonstrate that an open, community-driven personal AI agent can compete with — and in many ways surpass — anything a well-funded corporate lab has shipped.

What this tells us about what's coming

We work with AI every day, and OpenClaw represents a shift we've been watching closely: the move from AI as a service to AI as infrastructure you own.

The trajectory is clear. Models are becoming commodities. The value is shifting to the orchestration layer — how you connect AI to your life, your tools, your workflows. OpenClaw understood this before most of the industry caught up.

It runs on your hardware. It works from your chat apps. It remembers you. It grows with you. And because it's open source, it can never be taken away, paywalled, or enshittified.

The lobster has molted into its final form. And we think it's worth paying attention to.

---

Get started with OpenClaw: openclaw.ai

Star on GitHub: github.com/openclaw/openclaw