Policy Library

Upload PDFs or DOCX files — each document is chunked, embedded, and stored in the vector database for AI-assisted QA scoring.

How it works

After upload, each document is automatically split into 400-token chunks with 50-token overlap, embedded with OpenAI text-embedding-3-small, and stored in pgvector. When a call is processed, the top 8 most relevant chunks are retrieved and provided to both AI auditors.