Psychiatry relies on narratives, observation and interpersonal data more than definitive tests. Errors in inference can lead to inappropriate treatment, avoidable adverse effects and delayed recovery. This article proposes an investigative framework for clinical reasoning tailored to community mental health services, in-patient wards and liaison teams by translating hypothetico-deductive reasoning into clinical practice: broad-then-focused enquiry, explicit competing hypotheses, a reliability hierarchy for weighing clues, proportionate action under uncertainty, and time-bound review. The framework is positioned alongside existing approaches and considers the emerging role of artificial intelligence as a complement to clinician-led formulation. With fictitious vignettes, diagrams and summary tables, the proposed framework aims to improve formulation quality, support safer decisions before diagnostic closure, and make reasoning transparent to patients and teams. As a proposal, it requires empirical evaluation against routine practice in clinical and educational settings before any claim of effectiveness can be made.