Hierarchical frameworks of big ideas provide an important basis for organizing unit-based geography instruction and supporting the development of students’geographical competencies. Constructing such frameworks, however, places substantial demands on teachers’ability to synthesize relevant literature and make discipline-informed judgments. Drawing on a discipline-informed human–AI co-development approach, this study developed an AI-based workbench for generating hierarchical frameworks of big ideas in geography. The workbench decomposes the task into fi ve sequential processing nodes and integrates AI generation with teacher judgment through material routing, node-based generation, teacher review, and feedback-driven revision. This design creates a human–AI workfl ow in which intermediate outputs can be retained, revised, and traced. The workbench was fi rst tested on the topic of Industrial Location Choice, with the human–AI revision process documented in detail. Nine experts then independently evaluated automatically generated frameworks for seven geography topics. The overall mean rating was signifi cantly higher than the prespecifi ed benchmark of 3, corresponding to“basically meets the requirements”(M = 3.71, p < .01), and 92.4% of all ratings met or exceeded that benchmark. Directed content analysis identifi ed four areas in which teacher intervention was most needed: the selection and placement of core concepts, terminological precision, alignment across hierarchical levels, and connections to teaching practice. The fi ndings suggest that the workbench can support teachers in constructing hierarchical frameworks of big ideas in geography, but its outputs still require discipline-informed teacher judgment.