One intelligent foundation, four native capabilities, data-driven decision support across multiple workflows.
StarPlex is a pharmaceutical AI agent platform independently developed by Weiming Penguin. It deeply integrates four core AI technologies—knowledge graphs, dynamic ontology modeling, large-model cognition, and multi-agent collaboration—to create an AI-native solution spanning data collection → real-time cleansing → decision support → business execution, delivering intelligent data capabilities across multiple workflows for the pharmaceutical industry.
Four products. One shared semantic framework.
Take control of your data at the source
Intelligent multimodal data collection covers 12 types of heterogeneous data sources, including direct DDI connections, shipments, traceability codes, distribution flow data, e-commerce, DTP, and O2O, with AI-driven real-time collection and governance. Collect data without stationing staff on-site.
Significantly improve cleansing accuracy
The Tiangong™ pharmaceutical matching engine continues to advance through six generations, from “Manual · Zero Start” to “Tiangong · Sixfold,” improving cleansing rates and accuracy with morphological matching, dynamic scoring, pharmaceutical small models, LLM Embedding, and RAG Rerank.
Turn data from a “silent asset” into productive value
An enterprise-grade data asset workspace that consolidates pharmaceutical master data, with broad coverage of HCOs, HCPs, and scholarly literature resources. Build an intelligent closed loop of “data → analysis → prediction → decisions → action.”
An enterprise-grade, AI-native operational hub
Four-layer architecture: AI data foundation → AI monitoring and alerts → AI analysis and decision support → AI-automated execution. Integrates GeoAI, GNN, NL2SQL, generative AI, and multimodal perception.
Note: The product metrics and capability descriptions above are general introductions to functionality. Actual data scale and results vary depending on data conditions, deployment environments, and business scenarios.