Project summary
Global tätiger Pharmakonzern
Branche: Pharmazeutische Gesundheitsbranche
Projektstart: 04/2023
Schwerpunkt: Generative & Agentic AI
Use Case: Agentic Process Automation
Projektziel
Entwicklung einer Agentic-AI-Lösung, die Markt- und Wettbewerbsinformationen automatisiert analysiert und aktuelle Insights für fundierte Entscheidungen bereitstellt.
Wichtigste Kernfunktionen
- Vollautomatisierte Datensammlung aus verifizierten externen Quellen
- Anreicherung nach Kategorien, Sentiment, Trends und Auffälligkeiten
- LLM-gestützter Chatbot für Ad-hoc-Abfragen und explorative Analysen
Tech Stack
The Starting Point
Specialized departments within a global pharmaceutical company had to regularly analyze market and competitive information from news articles, press releases, and other reliable sources. Relevant topics included, among others, research pipelines, new drugs, acquisitions, investments, location developments, and macroeconomic indicators.
The existing process was largely manual: information had to be collected, summarized, evaluated, and consolidated for management reports. This took several workdays per report and made it difficult to respond promptly to new developments in the market.
The HMS Solution
Agentic AI for Automated Competitive Analysis
HMS developed a competitive intelligence tool that automates the research, analysis, and preparation of competitive information. The solution uses large language models to extract, summarize, and categorize relevant content from selected and qualified web sources.
An Agentic AI workflow orchestrates the analysis processes, manages LLM calls, and prepares the results for reports and interactive queries. This provides business departments with up-to-date, structured information without having to manually evaluate large amounts of data.
The solution processes news and press releases from reputable sources as well as news aggregators. Additional data sources can be integrated depending on the use case.
The platform uses selected data sources, structured analysis processes, and large language models to categorize and summarize information and identify trends or sentiments.
Yes. Using a built-in chatbot, users can interactively search through the processed information, ask ad hoc questions, and perform further analyses—without any prior technical knowledge.
The solution is based on Agentic AI and utilizes, among other things, Azure OpenAI, LangChain, AWS, and Palantir Foundry. Thanks to its modular architecture, models and data sources can be flexibly expanded.
