Case Study: Roche achieves faster, higher-quality patent research with Dataiku

Roche Boosting Case Law Analysis With Generative AI

Roche’s patent team faced a labor-intensive, high-stakes research process, spending many hours on legal analysis across roughly 50,000 European Patent Office appeal cases. To modernize this work, Roche adopted Dataiku and used it to bring GenAI into patent law analysis, starting with pilots for semantic search, full-text analysis, and deep search.

With Dataiku, Roche built Themis PatAI, an agentic AI workspace that routes legal and knowledge queries to the right tools in one interface and supports attorneys with summaries, follow-up answers, and document checks. The results included new GenAI projects built in months instead of months-long cycles, estimated annual savings of $100K–$250K from reduced attorney hours, and another $375K–$475K saved by avoiding consultancy costs, while also improving insight quality and attorney productivity.

Roche

Andres Buser

Chapter Lead New Modality IP