OpenClaw, the free and open-source AI agent platform that runs locally on your computer, has released version 2.0, marking the largest update in the project's history since its launch in November of last year. The update was built by 933 contributors, including 569 first-time contributors, and includes more than 16,000 pull requests. The project took nearly seven weeks to complete this release, a significant departure from its usual one or two-day patches.

The timing of the announcement drew attention for its playful competitive spirit. OpenClaw joked on X that it had beaten Grand Theft Auto 6, the video game scheduled to release in November 2026. While the comparison was tongue-in-cheek, it underscores how rapidly the open-source AI agent ecosystem is evolving and how seriously the community takes its development pace.

What's Actually New in OpenClaw 2.0?

The update focuses on making the platform more accessible and easier to use, particularly for people new to AI agents. One of the most significant changes is simplified installation. Instead of requiring users to configure everything from scratch, OpenClaw 2.0 now detects and integrates with tools already on your computer, including existing ChatGPT, Claude, or Codex logins, API keys, and local models such as Ollama or LM Studio.

The browser app, which the project identified as the primary entry point for most users, has been completely rebuilt. It now opens directly into a conversation with your AI agent, called a Claw, and is designed to streamline both initial setup and ongoing work. The new interface centers conversations with chat, files, approvals, and settings positioned closer together for easier navigation.

Beyond the interface, OpenClaw 2.0 introduces several practical features that expand what users can accomplish:

• Pinned Chats and Grouped Sessions: Users can now organize conversations more effectively, with the ability to pin frequently used chats and group related sessions together for better workflow management.
• Live Markdown Streaming: The platform now supports real-time formatting as the AI agent generates responses, making output more readable and professional.
• Side-Conversation Command: A new "/btw" command allows users to ask tangential questions without losing the context of their main conversation.
• Shared Cloud Sessions: Team members can now join live work sessions and collaborate on tasks while maintaining full context, a feature that previously required losing the AI agent's accumulated knowledge when handing off work.

How to Get Started With OpenClaw 2.0

• Mac, Linux, and Windows Installation: Download the new version and let it detect your existing API keys and logins. The setup process now prioritizes getting you to your first conversation quickly, with configuration options available later through chatbot interaction.
• Mobile Setup on iOS and Android: The onboarding process on mobile devices now begins with pairing and permissions, streamlining the initial connection process for smartphone and tablet users.
• Data Migration: If you're upgrading from an earlier version, note that session and transcript storage has moved from individual files to SQLite format. Back up your legacy data before migrating, as sessions created after the migration will not appear in older versions.

What Can You Actually Do With OpenClaw?

The platform is designed to handle workflows that start simple and grow in complexity without becoming harder to manage. For example, a user could set up a Claw to watch their email inbox for school-related messages from their children and send a Telegram notification when something important appears, such as homework assignments or upcoming activities.

More sophisticated workflows are also possible. A user could ask their Claw to find information from an email receipt when a family member sends an iMessage asking which iPad was purchased for their father, and have the Claw automatically send the answer back. These multi-step tasks that span different apps and data sources represent the kind of practical automation that OpenClaw aims to make accessible to non-technical users.

The update also introduces new collaboration features. Conversations now support identity, permissions, and presence indicators, allowing multiple people to work together on the same task. An optional incognito mode is available for sensitive work. Settings have been reorganized into a full-page workspace for easier customization.

The Community Behind the Update

The scale of this release reflects the growing momentum in the open-source AI agent space. The project previously released 106 updates in just 230 days, demonstrating a rapid iteration cycle. However, OpenClaw 2.0 represents a deliberate shift toward larger, more comprehensive updates. The involvement of 569 first-time contributors suggests that the project is attracting new developers and expanding its community base.

It's worth noting that Peter Steinberger, OpenClaw's creator, joined OpenAI in February 2026. Despite this, the project continues to be supported by the independent OpenClaw foundation and its community of contributors, ensuring that development remains community-driven rather than corporate-controlled.

For users considering upgrading, the migration process includes important safeguards. Migrations from Claude, Codex, or Hermes now run in a staging area with verification before the new setup replaces the old one, reducing the risk of data loss or configuration errors during the transition.