The OpenClaw Foundation has launched OpenClaw 2.0, marking the open-source AI platform’s largest update to date with over 16,000 pull requests integrated. The release introduces a streamlined onboarding system that automatically detects a host machine’s existing API keys, local models, and subscriptions for services like ChatGPT and Claude, reducing initial setup steps.
The update features a rebuilt browser application containing a “Session Rail” interface for real-time task tracking, alongside a companion thread that allows users to query running sessions via a separate utility model. Additionally, OpenClaw 2.0 introduces Shared Cloud Sessions, allowing multiple users to collaborate on ongoing agent tasks and context. Credentials remain stored locally on the gateway and are not passed to remote hardware.
OpenClaw 2.0 supports three execution environments: local gateways, user-owned hardware via paired devices, and disposable remote machines provisioned through its Crabbox tool on backends like AWS and Hetzner. The update also includes modifications to memory, automations, security, and native application support, with code made available publicly on GitHub.
Why it matters
Open-source agent infrastructure is maturing, enabling multi-user cloud collaboration without exposing local API keys to remote workers.
Automated environment detection simplifies local developer setup, lowering onboarding friction for open-source AI toolkits.
Flexible execution environments allow operators to seamlessly scale execution from local desktop setups to cloud providers like AWS.
Source: the-decoder.com
