A major challenge for pervasive computing is to support continuous adaptation of applications to the behavior of the user. Recent research has adopted classical workflows as alternative programming paradigm for pervasive applications and approaches for context aware workflow models have been presented. However the current approaches suffer from the low flexibility of classical workflow models. We present a solution that allows attaching workflows to real-world objects and defining relevant context dynamically in relation to those objects. The benefits are a dynamic, yet simple modeling of context constraints and events in pervasive workflows and a greatly reduced amount of context information that must be provided to the workflow.