A weekend project just changed where the entire AI industry is heading.

Peter Steinberger created OpenClaw in November 2025. By early February, it had over 100,000 GitHub stars, two million visitors in a single week, and multiple AI labs fighting over who'd get the creator. OpenAI won.

The open-source project is moving to a foundation. Steinberger joins OpenAI to build the consumer version — personal AI agents for everyone, not just developers.

The most capable AI lab in the world looked at a self-hosted, open-source agent platform and said: this is what we need to bring to a billion people. That's not an acquisition. That's a signal.

The MVP that proved the thesis

OpenClaw isn't a chatbot. It's scaffolding that lets AI models actually do things — read your calendar, draft your emails, manage your files, coordinate across your apps. The AI doesn't just talk to you. It works for you.

And it works. Over 3,000 community-built skills on ClawHub, its marketplace. People weren't just chatting with their agents — they were building entire workflows. Research assistants that scan feeds and produce morning briefings. Writing teams that draft, fact-check, and edit. Administrative agents that manage scheduling, follow-ups, and document preparation.

A solo developer with a side project just validated what every major lab suspected — people want agents, not chatbots. They want AI that does things.

Every lab wants the agent layer

The acqui-hire is a signal, not a starting gun — labs were already building this. But OpenClaw proved the demand is real, right now:

  • OpenAI hired Steinberger and is backing the foundation. A consumer-facing agent platform is coming — likely integrated into ChatGPT.
  • Anthropic has been pushing Claude computer use and launched Claude Cowork in January — persistent agents that work alongside you in your projects.
  • Google is building agent capabilities across Gemini, with deep integration into Workspace, Android, and their cloud platform.
  • Microsoft is weaving Copilot agents into every corner of the Office ecosystem.

Chatbots were the demo. Agents are the product. The competition now is for the layer between the AI model and your actual life — your calendar, your inbox, your workflows, your decisions.

Whoever makes this feel effortless for ordinary people wins.

Team in a box

So what does this actually look like for a business?

Say you're a marketing executive at a mid-sized company. You've got a team of five or six: an analyst, a content writer, a social media manager, maybe a designer and a front-end developer. Each costs £40–70K a year. Months to hire. Weeks to onboard. Years to get working well together.

Now imagine a preconfigured AI agent team — a "team in a box" — purpose-built for marketing. An analytics agent that monitors your campaigns and surfaces insights every morning. A content agent that drafts articles in your brand voice. A social media agent that schedules, engages, and reports. A design agent that generates assets on demand. A developer agent that builds landing pages and email templates.

Not replacing your human team. Amplifying them — giving each person the output capacity of three or four. Or giving a solo operator the capabilities of a full department.

This is the Viking ship principle. Small crews. Everyone creates value. No passengers.

AI doesn't just let businesses stay lean for longer — it removes the ceiling entirely. A five-person company with the right agent infrastructure can operate with the capability of a fifty-person one. And this isn't theoretical. People are already running agent teams that handle research, writing, editing, SEO analysis, and publishing — end to end. The tools exist. They're just not accessible to everyone yet.

That's what OpenAI is about to change.

A few things worth watching

Platform dependency. If you build your business workflows around one agent platform and it pivots, stalls, or gets absorbed — what's your migration path? The WhatsApp precedent is worth remembering: Facebook promised WhatsApp data would stay separate. Within two years, they'd merged it. The EU fined them €110 million.

Foundation governance. Steinberger's commitment to keeping OpenClaw open seems genuine — "a place for thinkers, hackers and people that want a way to own their data." But foundations can be shaped by their sponsors. OpenAI "continuing to support" the project means OpenAI has a seat at the table. Probably the biggest seat.

The LLM pipeline. OpenClaw runs locally — your data stays on your server, which is the right architecture. But every prompt still flows through an LLM provider. If OpenAI becomes the default recommended provider for the platform they're sponsoring, that's a concentration of data worth thinking about.

Not reasons to hold off. Just reasons to pay attention.

What to do about personal AI agents

If you're running a business — especially a small or mid-sized one — personal AI agents are about to become the most significant productivity tool since the smartphone.

Start building now. Don't wait for the polished consumer product. OpenClaw is open source and free. Anthropic's Claude Cowork is live. The learning curve is real, but the people who start now will have a serious head start on everyone who waits for it to be "easy."

This is especially relevant for UK businesses, where the problem isn't AI — it's growth ambition.

Think in teams, not tools. The unlock isn't a single AI assistant — it's a coordinated team of agents, each with a specific role. Researcher, writer, analyst, developer. The "team in a box" model works because it mirrors how human teams operate, but without the coordination overhead.

Own your infrastructure. Self-hosted and open-source means your data, your configurations, your agent's accumulated knowledge — all on your hardware. That matters when platforms get acquired, pivot, or change their terms.

Keep your options open. Pay attention to which LLM providers your workflows depend on and how easy it would be to switch. The market is still forming.

Every major lab is now racing to build the personal agent layer. Your competitors won’t wait for you to catch up.