What does it take to run the fastest-moving open-source agent platform as something you actually depend on, and then adapt it for environments that require stronger isolation, safety, and governance? In this two-part session, an OpenClaw maintainer will share firsthand what dogfooding the project as an always-on agent looks like: keeping pace with a codebase that ships daily, upgrading without losing state, and turning the failures found in a real homelab and family production environment into upstream fixes. Then, we'll explore how OpenShell, organizational policies, and configuration controls can be layered onto upstream OpenClaw for more controlled deployments. Together, the speakers will examine the boundary between the upstream project and deployment-specific requirements. They'll share lessons relevant to anyone bringing open-source software into the enterprise.