Formulation development remains a knowledge-intensive activity in which scientific judgment is distributed across protocols, excipient knowledge, experimental records, and dissolution interpretation. AI assistance could help organize this complexity by connecting evidence sources that are normally reviewed separately. Formulators often move manually between protocol folders, spreadsheet-based excipient records, and dissolution analysis tools. This fragmented workflow can make it difficult to identify relevant precedents, compare similar formulations, and explain why a formulation failed to meet a release target. A multimodal AI copilot could ingest internal development protocols, excipient property databases, and dissolution curves to support natural-language formulation queries. Such a system would not replace the scientist but would help retrieve evidence, suggest formulation adjustments, and generate rationale for expert review. The proposed copilot includes a document-retrieval module for protocols and development reports, an excipient-property knowledge graph, a dissolution-curve encoder, a multimodal reasoning engine, and a conversational interface. Together, these modules would allow the system to connect text, structured formulation attributes, and release-profile behavior. By providing traceable, evidence-based responses, the copilot would be expected to reduce cognitive load during formulation design and troubleshooting. It could also help preserve institutional knowledge by turning historical development experience into searchable, reusable evidence. A formulation AI copilot could support a shift from experience-based trial-and-error toward data-driven, hypothesis-guided formulation development. Its value would depend on careful validation, human oversight, and integration into regulated pharmaceutical workflows.