Industrial multimodal demo for R&D
- 6demo acts
- 3languages: ES, EN, IT
The problem
R&D teams work over large, heterogeneous technical corpora where what matters is buried in figures, tables and documents in several languages.
The architecture
A demo structured in 6 acts over a technical corpus in PDF and XLSM, with interaction in English, Spanish and Italian.
It combines advanced search, document processing, multimodal understanding and agents on Azure AI Foundry. Not a generic chatbot: every answer keeps the link back to its source document.
Impact
- Dozens of technical documents with multiple tables and curves, made navigable
- Understanding of figures and tables, not just plain text
- Interaction in English, Spanish and Italian over the same corpus
- Document traceability in every answer
Stack
- Azure AI Foundry
- Multimodal
- Azure AI Search
- Multi-Agent
- Python