This paper describes and benchmarks the competitor discovery components used within an agent-based AI system for rapid pharmaceutical asset due diligence. Given a specific indication, the competitor discovery AI agent searches for all drugs that comprise the competitive landscape for that indication and extracts standardized properties of these drugs. Competitor definitions vary across investors, data is paid/licensed, distributed across multiple registries, has inconsistent ontologies across indications, has numerous aliases, is multimodal, and rapidly evolving. Existing LLM-based AI systems cannot reliably search all competitor drug names, and there is no public benchmark for this task. To address this, we transformed five years of multimodal unstructured due diligence notes from a private biotech VC fund into a structured valuation corpus to map competitor drugs by indication and standardized properties. Furthermore, we introduced a competitor validation LLM-as-a-judge agent to eliminate false positives, improve accuracy, and suppress hallucinations. The competitor discovery agent presented in this paper achieved 83% recall, outperforming OpenAI Deep Research (65%) and Perplexity Labs (60%). This system is deployed for corporate users, and in a case study at a biotech VC investment fund, analyst processing time for competitive analysis was reduced from 2.5 days to approximately 3 hours (approximately 20x).
