PharmaRadar

Pharmaceutical markets evolve continuously through regulatory actions, product launches, and competitive communications, yet most market intelligence processes remain manual, delayed, and fragmented. We were inspired by the growing gap between how quickly the market changes and how slowly insights reach decision-makers. This led us to explore whether an always-on, AI-driven system could continuously listen to external pharma signals and transform them into timely, contextual intelligence rather than static reports.

PharmaRadar is designed as an always-on market intelligence system that continuously monitors public pharmaceutical data sources and converts unstructured external information into decision-ready insights. The platform detects competitor activity, product launches, communication shifts, and high-impact market events by building long-term market memory and reasoning over changes in context. Instead of relying on keyword-based alerts, it delivers prioritized, context-aware insights grounded in historical evidence through a centralized dashboard and alerting interface.

The system was built as a modular and scalable intelligence pipeline. Automated workflows created with n8n ingest data from pharma brand websites, regulatory announcements, and medical news sources. A FastAPI backend cleans and enriches this data using entity extraction and classification techniques. Semantic embeddings are generated and stored in a vector database to maintain persistent market memory, while a Retrieval-Augmented Generation pipeline powered by the Gemini API retrieves relevant context and generates grounded insights. These insights are surfaced through a Streamlit-based dashboard designed for clarity and decision focus.

One of the primary challenges involved handling highly unstructured and inconsistent data across diverse sources while ensuring retrieval remained relevant and noise-free. Designing a RAG pipeline that produced explainable, grounded insights without introducing hallucinations required careful tuning. Balancing automation, interpretability, and reliability proved to be a critical challenge throughout development.

Within the hackathon timeframe, we successfully delivered an end-to-end, always-on intelligence system that integrates automation, semantic memory, and contextual AI reasoning. We are particularly proud of building a platform that goes beyond surface-level monitoring to detect strategic shifts and market signals while maintaining transparency and explainability suitable for enterprise decision-making.

This project reinforced the importance of combining continuous automation with historical context rather than relying solely on real-time data. We learned that market intelligence becomes significantly more valuable when insights are grounded in long-term patterns and semantic relationships, and we gained practical experience designing RAG-based systems that balance scalability, accuracy, and interpretability.

Looking ahead, PharmaRadar can be extended to include broader medical and social discourse analysis, predictive trend modeling, and multi-language support. Integrations with collaboration tools such as Slack or email, along with more refined impact scoring, would further enhance its ability to support proactive regulatory, competitive, and strategic decision-making in pharmaceutical organizations.