This chapter examines the possibilities and challenges of augmenting expert reasoning in complex, high-risk work settings like healthcare, the judicial system, engineering, military defense, public safety, and more. While AI-augmented reasoning has immense potential in such domains, there is no clear developmental pathway for how such tools should be designed. AI is problematic because it does not follow a human reasoning process nor collaborate as humans do. It can lack transparency and consistency, and its inconsistent outputs can cause unreasonable errors. These characteristics of AI make it potentially incompatible with human experts and the accountabilities of their domain. Creating reliable methods to integrate AI and human participation in reasoning tasks remains an open question. Toward this aim, this chapter presents six factors as foundational considerations for studying and designing tools for AI-augmented human reasoning. The factors include contextual adaptation; situated accountability; roles, boundaries, and interdependencies; domain-specific reasoning strategies; new interaction modes; and shared knowledge and values.