The OpenClaw Foundation has released OpenClaw 2.0, a major update to its open-source AI platform. The release is described as the project's largest so far, with over 16,000 pull requests and a set of changes aimed at making the software easier to start, easier to monitor, and easier to use with other people.
The central story is not a single interface change or one new deployment option. OpenClaw 2.0 combines a simpler first-run experience, a browser app rebuilt from scratch, and Shared Cloud Sessions that let multiple users work on the same tasks with context attached.
Setup now starts with what users already have
The biggest change in OpenClaw 2.0 is the simplified first-run setup. Instead of asking users to work through many configuration steps at the beginning, the software now detects existing resources on the computer.
Those resources include ChatGPT or Claude subscriptions, API keys, and local models. That matters because OpenClaw is positioned as a platform that can work across different model access paths rather than forcing every user through one narrow setup flow.
The practical effect is a shorter path from installation to use. Users who already have subscriptions, keys, or local models can let OpenClaw identify those pieces and move forward with fewer manual steps.
Configuration has not disappeared. The release changes when and how more settings can be handled. Additional settings can be adjusted later during a conversation with the bot, which moves setup from a front-loaded process into the flow of actual work.
The browser app is built around active work
OpenClaw 2.0 also brings a browser app rebuilt from scratch. The new browser experience opens a conversation right away, making the chat the starting point for continuing setup, resuming ongoing work, or watching tasks as they happen.
One of the notable additions is the "Session Rail." It is a compact status display inside the chat that shows the progress of a running session. The source article lists ratings, plan progress, and pull requests as items that can appear there.
That makes the browser app more than a place to type prompts. It becomes a workspace for following what the AI session is doing, what stage it is in, and what work products are connected to it.
The app also includes an integrated companion thread. This lets a user ask questions about the current session without interrupting the agent. According to the documentation cited in the source article, the gateway loads a limited snapshot of the session for this feature and uses a separate utility model.
The distinction is important within the facts given: the main session can keep running, while the user still has a way to inspect or clarify what is happening. That can reduce the need to choose between waiting silently and disrupting the agent's current task.
Shared Cloud Sessions add collaboration
Shared Cloud Sessions are another major part of OpenClaw 2.0. They let multiple users work on the same tasks, bring team members into ongoing work, or assign tasks together with their context.
For teams, the context piece is the key detail. A task in progress can carry surrounding information with it, which makes collaboration less dependent on one person manually explaining the state of the work.
OpenClaw says its own team already uses Shared Cloud Sessions to build OpenClaw itself. The source does not add more detail about that internal use, but it does show that the feature is not presented only as a future-facing concept.
The release also gives users several places where sessions can run:
- On the local gateway, which is the default.
- On the user's own hardware connected through openclaw connect, described as Paired Devices.
- On rented disposable machines through the Crabbox provisioning tool.
Crabbox supports backends like AWS and Hetzner. According to the documentation cited in the source article, provider credentials always remain on the gateway and never reach the remote machine.
That detail helps define the architecture described in the release. OpenClaw 2.0 is not just adding cloud execution in a generic sense; it separates where credentials are kept from where a remote session may run.
A larger release beyond three headline features
The three most visible changes are setup, the browser app, and Shared Cloud Sessions. But the release notes also list updates across messaging, memory, skills, model support, automations, native apps, plugins, and security.
The source article does not describe those areas in detail, so the safest reading is that OpenClaw 2.0 is a broad platform release rather than a narrow interface update. The headline features make the platform easier to enter, easier to supervise, and easier to use collaboratively.
For users evaluating OpenClaw 2.0, the practical questions are straightforward. Do they want setup that can detect existing subscriptions, API keys, and local models? Do they need a browser app that opens directly into a conversation and shows live session status? Do they want multiple users to work around the same AI-driven task with shared context?
OpenClaw 2.0 answers those questions with a release built around workflow continuity. The update reduces the friction at the beginning, gives users more visibility while work is running, and expands sessions from individual use into shared cloud-based collaboration.