Skip to content

Financial Research RAG over 10-Ks and Earnings

A finance-grade RAG stack that answers analyst questions across 10-Ks, transcripts, and internal notes with inline citations.

LlamaIndexOpenAI APIFastAPIPostgreSQL
Abstract indigo and amber cover illustration for the Financial Research RAG project, showing floating financial-table planes with amber citation beams.

Highlights

  • Table-aware chunking for financial statements and MD&A sections.
  • Time-scoped retrieval so answers respect reporting periods.
  • Citation-first synthesis with refusal on stale or thin context.

Outcomes

  • Cut analyst prep time per company by 70% during earnings season.
  • Zero hallucinated numeric claims across 5k audited responses.

Stack

  • LlamaIndex
  • OpenAI API
  • FastAPI
  • PostgreSQL
  • Pinecone