The assistant functions by mapping user inquiries to a validated knowledge space, moving beyond simple information retrieval to execute authorized system actions. By automating repetitive tasks, the tool aims to shift focus back to patient care and complex judgment calls. Each action taken by the AI requires user confirmation and remains tethered to existing permissions and audit trails to meet strict regulatory standards.

Early adopters, including Daiichi Sankyo, have integrated the assistant into daily workflows to streamline document summarization and system navigation. Unlike general-purpose models, the system operates under a zero-data retention policy and avoids using customer information to train public models. The assistant currently supports over 40 distinct topics, with capabilities expanding to match the specific needs of sponsors and research organizations.