AI Figure Editor is a scientific figure editor for researchers, authors, and research teams who need to revise schematics, study designs, pathway diagrams, model architectures, and graphical abstracts. It lets users upload an existing figure, identify the area or content to change, and describe a revision in natural language while keeping the original available for comparison. Users can mark regions with notes, rename labels, reroute connections, rebuild visual hierarchy, normalize colors and line styles, recompose panels, and reduce diagram density without losing specified entities or relationships. Guided workflows support methodology diagrams, biological pathways, longitudinal study designs, and AI or machine learning architectures. Each request creates a new version in the revision conversation, and users can continue from a selected result or turn reviewer and PI feedback into a concrete revision plan. The workflow preserves source context and makes requested changes traceable, helping authors compare drafts and verify scientific meaning before publication. Unlike general-purpose image editors, it focuses on structured scientific figure revision with explicit preservation instructions and human review checkpoints.