OpenClaw 2.0 is now available. The biggest update in the brief but striking history of the AI tool opens, according to its creators, a new chapter. OpenClaw 2 has been made possible thanks to 933 contributors, 569 of them taking part for the first time, and more than 16,000 pull requests to the project. We’re looking at a huge leap in the scope and flexibility of a key tool for those of us who want AI to move from answering questions to helping us with our day-to-day tasks.
Simpler installation and a browser that takes center stage in the update
During installation, OpenClaw 2 detects any ChatGPT or Claude subscriptions we may have, our API keys, and the local models available. With very little extra setup, we can start chatting much sooner and even finish the rest of the setup while already chatting with Claw. An approach we’ve already seen in other tools, including ChatGPT’s integration with Mac apps.
After the initial setup, the main big new feature is the browser app. From there we can chat, continue with tasks we already have underway, and track the progress. Much like Claude’s computer control feature, the most interesting thing is that here we start with a simple task or automation and build on it.
We can, for example, keep an eye on our emails, retrieve information from our services, or reply to the messages that reach us through different channels. And here comes the third big new feature, because in OpenClaw 2.0 sessions let us bring in other people and share our tasks while preserving the context.
It’s an approach that fits with multi-agent systems like OpenAI’s Astra, where multiple participants work together, in this case humans or AI agents, to complete a task. A totally new way of working, and one we’ll want to explore, given its enormous potential, in great depth.
