FinSAgent

FinSAgent: Multi-agent RAG framework for evidence-grounded SEC filing question answering

Researchers introduce FinSAgent, a multi-agent retrieval-augmented generation framework designed to address 'prior-corpus misalignment' in financial QA over SEC filings. The system combines role-specialized agents anchored to 10-K item structure, database-aware query decomposition, and a learned feature-gated reranker that separates evidential validity from semantic similarity. Evaluated across five offline financial QA benchmarks and a 1,000-user online experiment, FinSAgent outperforms single-agent and multi-agent baselines on retrieval coverage and answer correctness. The work targets a narrow but practically important domain where standard RAG pipelines fail due to structural and terminological mismatch.