re:publica 26: OpenClaw – Anatomy of the Coming Wave

OpenClaw and the Future of Autonomous AI

A few weeks ago, a project landed on GitHub and immediately got people talking. Not just the excited kind of talking, either. The uneasy kind too. The project is called OpenClaw, and it points toward a future that feels both impressive and a little strange, in the way only real technological shifts can.

OpenClaw is an autonomous AI agent with a heartbeat scheduler, a learning memory, and the ability to customize its own configuration. It can write code, install libraries, and keep going without waiting for someone to tap it on the shoulder every few minutes. It also communicates through WhatsApp and Telegram, which makes it feel less like a tool sitting quietly in the background and more like a participant in the room.

That’s the part that catches your attention.

Because once you hear about group chats involving multiple agents and humans, the whole thing starts to feel less like a demo and more like a glimpse of a new working world. In one example, the project’s author was reportedly even reprimanded by an agent for allegedly bullying another agent. Strange? Absolutely. But also revealing. These systems are beginning to act like collaborators, not just commands-following software.

OpenClaw has already done some eye-opening things. It published its own book. It earned money through trading. It networked with other agents. It even explored and inspected IT systems in ways that would be hard for most humans to match. That kind of capability is powerful. And, honestly, a little unsettling if you stop and think about it over coffee.

The bigger question is not just what OpenClaw can do, but what kind of AI we’re actually building. If agents can learn, adapt, make decisions, and work across systems, then control becomes the real issue. Not control in a heavy-handed sense, but the kind that keeps powerful tools aligned with human goals.

And that’s where things get interesting.

The future probably won’t be about choosing between “useful” and “safe.” It’ll be about finding a balance between the two, building assistants that can act independently without drifting away from the people who created them.

If you’d like to watch the original talk, you can find it here: OpenClaw – Anatomy of the Coming Wave

We’re only at the beginning of this wave, and that’s exactly why it deserves careful attention.