SpainHealthcare and life sciences2025 – ongoingAlmirall
R&D document search assistant on Azure OpenAI
Almirall's scientists searched 400,000 R&D documents in seconds instead of hours or days with an Azure OpenAI assistant
Problem
Almirall's research teams held more than 50 years of R&D documents in English, Spanish and Catalan. Finding data in them could take hours or days.
Approach
Almirall built a custom assistant on Azure OpenAI in Foundry Models, Azure AI Search and Azure Databricks. Scientists ask questions and search the documents in all three languages. It was rolled out to teams in R&D, and the company wants to extend it to other departments. Separately, staff use Microsoft 365 Copilot for tasks such as updating the R&D governance handbook, summarising documents and building presentations.
Results
Takeaway
Users judged the assistant accurate about 80% of the time, so answers still needed checking by scientists.
Categories
- Industry
- Healthcare and life sciences
- Business function
- IT and R&DInternal work and knowledge management
- Innovation type
- Process innovation
- AI technology
- Generative AI
- Adoption
- Partner co-development
- Evidence
- Vendor case study
Sources
Source summaries
Microsoft's customer story describes a custom R&D search assistant at Almirall covering 400,000 documents in three languages. Searches that took hours or days now take seconds, and users report accurate answers about 80% of the time. It is in use in R&D and planned for other departments.
Full details are in the original sources.
