Mainstream digital graphic design tools seldom evolved since their creation, more than 25 years ago. In this dissertation, I address the following questions: How do designers work with design software? And how can we design novel design tools that better support designer practices? Using StoryPortraits, a method designed to capture rich qualitative insight, I first study four designer practices, ranging from specific design operations such as color selection, alignment and distribution, to more complex endeavors such as layout structuring and collaboration with developers. In these empirical studies, I characterize the existing mismatch between current digital design tools and designers practices. I show how design tools, because they decouple creativity from tool use, prioritize values such as efficiency and user-friendliness. Based on my empirical findings, I propose a new type of design tools, Graphical Substrates that combine the strengths of both programming and traditional Graphical User Interfaces. I design nine different tools that address the needs identified in the four empirical studies by reifying specific user process into Graphical Substrates probes. In four structured observation studies, I show how designers can appropriate these probes in their own terms. In this thesis, I argue that Graphical Substrates open the design space of designers' tools by bridging the gap between programming and graphical user interface to better support the wealth of designers' practices.