What we imagine possible shapes what we design and build, and what gets designed and built shapes what we imagine possible. This recursive loop happens in part because to design and build any system, one must draw ontological boundaries -- boundaries that define the nature of what is. Every cycle of design and development that assumes the same boundaries reinforces them as a given rather than a choice, making other boundaries and the worlds in which they exist difficult to imagine. As AI systems proliferate, this limitation becomes more consequential because their embedded assumptions risk foreclosing alternative futures before they can be imagined. Although the discourse around narrowing of possibilities is not new, systematic methods for translating such critiques into practice remain elusive. This gap persists in part because critique typically occurs after systems are deployed, leading to patchwork fixes that address symptoms without addressing the foundational assumptions that frame the problem definition. Combining critical analysis with practice, demonstrate that revealing and questioning the taken-for-granted boundaries through analytical frameworks, technical and design practice, and participatory approaches expands the space of imaginable possibilities and creates the conditions for ontological multiplicity in AI and computing development.