Skip to main content

​
The problem

Company knowledge lives in PDFs, decks, spreadsheets, and wikis nobody can search across. New hires ask the same questions in Slack for months.

​
How Ragrails solves it

Parse everything into one private, self-hosted collection. No documents leave your infrastructure.
  1. Parse folders of mixed file types, or accept uploads as bytes (no disk write).
  2. Store in a vector database you host.
  3. Retrieve or chat privately.
from ragrails import RagRails

rag = RagRails()

# Parse a shared drive of mixed documents
rag.ingest(
    docs={"folder": "files/company/"},   # PDF, DOCX, XLSX, PPTX, MD, CSV...
    embedding={"provider": "voyage", "model": "voyage-3"},
    storage={"vector_db": "qdrant", "collection": "wiki", "url": "http://localhost:6333"},
)

# Ingest an upload straight from memory (e.g. a web app)
rag.parse(files=[{"content": uploaded_bytes, "filename": "handbook.pdf"}])
Uploads can be ingested as raw bytes. See parse input forms. Everything runs against your own vector database, so data stays in-house.

​
Features used

Document ingestion · Storing · Chat