From an AI chatbot to agentic research memory at Bayer
- Pharma
- R&D
Challenge
Bayer R&D needed hypotheses grounded in a large body of preclinical literature, with reasoning scientists could check — more than a chatbot could give.
Solution
Cognee turned the literature into a knowledge graph, combining graph and vector reasoning so agents connect new data to prior findings and keep their sources.
Results
- 10,000 scientific papers compressed into research memory
- Hypothesis generation from months to hours
- Explainable output scientists can trust
- Knowledge graph
- Vector search
- Provenance
