We built software to support student reasoning about a phenomenon and development of hypotheses to explain it. The goal is to engage students in asking questions, generating hypotheses and testing predictions. Rashi, an intelligent tutor, tracks students’ investigations (e.g., hypotheses, questions, data collection, and inferences) and helps articulate how evidence and theories are related. The tutor provides advice, such as recognizing when data does not support a hypothesis Cases are presented in geology, biology or engineering, and students are scaffolded to use an inquiry-based approach to posit a theory to explain the situation. Generic and reusable structured tools guide students through exploration of ill-structured problem spaces, supporting student knowledge and scaffolding reasoning and diagnostic skills.