AI-Assisted Research System
A RAG-based research assistant trained on investment research reports, economic papers, and regulatory publications. Analysts query a curated knowledge base in natural language and get cited answers in seconds.
Outcome
Reduced research cycle from days to hours. Analysts cover 4x more companies with the same team.
Live Demo
Example Queries
Principal adverse impact indicators require mandatory disclosure for Article 8 and 9 funds...
National competent authorities must verify fund classifications are substantiated...
Transition risk from carbon-intensive holdings represents the largest near-term exposure...
The anti-greenwashing rule applies to all regulated firms making sustainability claims...
Select a query above...
The Challenge
Investment research analysts were spending 60-70% of their time reading through historical reports, regulatory publications, and economic papers to find relevant context for current analysis. The knowledge existed in the organisation but was not accessible without manual search and reading.
How It Was Built
Built a RAG-based research assistant using LangChain, Pinecone vector store, and OpenAI. The system ingests investment research reports, economic papers, regulatory publications, and earnings transcripts, stores them as vector embeddings, and enables natural language queries with cited source retrieval. Analysts can ask complex analytical questions and receive grounded answers with document references, confidence signals, and source excerpts.
The Result
Research cycles that previously took 3-4 days of background reading can now be initialised in under an hour. The same team covers 4x more companies without adding headcount. The system particularly excels at cross-document synthesis: identifying where multiple sources agree or diverge on a theme.