The process of large-scale science must evolve to facilitate the next steps of scientific discovery. Grid technology and semantic tools will be valuable in dealing with the complex multidisciplinary simulation and data environments that next-generation science will require. We envision semantic Web-like tools that automatically check the validity of sequences of composed operations and data, and automatically construct intermediate steps in a loosely specified sequence. These tools should also automatically construct sequences of operations that are consistent with a discipline model representing permitted relationships among simulation and analysis operations and data for particular disciplines, such as climatology or high-energy physics. These tools are called semantic services.