Always-On Autonomous AI Agents: Exploring the OpenClaw Abstraction

As agents evolve from basic conversational interfaces to personal complex task navigators, the primary challenge shifts from writing context strategies to designing robust systems. This hands-on workshop explores the engineering needed to manage persistent agent state, secure tool execution, and complex workflow creation in a low-code, high-capacity context. Using OpenClaw, a popular open-source agent gateway, we illustrate how a single daemon can manage sessions across multiple applications, maintain long-term interpretable memory, and safely execute tools in a virtual user-centric environment. We emphasize low-code abstractions and out-of-the-box capabilities while maintaining enough rigor to motivate effective and secure best practices. You’ll be able to expand on these ideas and improve your agents’ capabilities as new innovations arrive and the field continues to evolve.

Workshop Preparation: - Please bring your own laptop. - Please review the following document and prepare accordingly before the workshop: https://developer.nvidia.com/dli/getready