Grid computing tasks are often broken down into multiple subtasks and connected using a directed acyclic graph (DAG) to form a grid workflow. If we assume a dynamic market-based environment with multiple virtual organizations competing to provide grid computing resources, it becomes apparent that a static mapping of these resources to the workflow will produce suboptimal results. Furthermore, the authors argue that a rigid workflow model will also produce suboptimal results within a dynamic environment. The multi-agent systems community addresses the problem of executing high-level tasks through a series of coordination mechanisms. If we view grid computing service providers and consumers as agents, we can view the problem of executing a grid computing task as coordination between agents. The position taken in this paper is that TAEMS, a domain independent model of the problem solving activities of an intelligent agent, can provide a flexible workflow model that incorporates the notion of quality for grid computing tasks. This flexible workflow approach improves the capabilities of workflow execution engines to optimize a workflow.