Abstract: Thrombotic diseases impose a substantial global burden, while traditional models fall short in tailoring care to heterogeneous patients. The rise of Generative AI(GenAI) enables a “human-GenAI coexistence” paradigm in which clinicians, patients, and GenAI collaborate as equal partners to deliver precision antithrombotic therapy through multimodal evidence integration, explainable individualized decision-making, and closed-loop interventions. This review proposes a three-subsystem architecture—Evidence, Decision, and Intervention. The Evidence subsystem transforms heterogeneous clinical and omics data into structured disease knowledge; the Decision subsystem employs multi-criteria analysis and patient-preference modeling to generate auditable treatment plans; the Intervention subsystem leverages GenAI’s multi-role design and a coordinated “nudge-boost” strategy to strengthen adherence and long-term self-management. Across use cases—intravenous thrombolysis, perioperative venous thromboembolism prevention, individualized antithrombotic selection, and mobile health—this paper illustrates gains in predictive performance, transparent shared decision-making, and dynamic optimization, while addressing challenges around hallucination, interpretability, liability, and standardization.This paradigm provides a replicable theoretical framework and implementation pathway for optimizing pharmacotherapy in thrombotic diseases and advancing patient-centered precision medicine.