Skip to main content

​
The problem

Documentation grows faster than anyone can navigate it. Users give up on search and file a ticket, or leave.

​
How Ragrails solves it

Crawl the whole docs site once, then answer questions grounded in the actual pages.
  1. Full-crawl the docs site into a collection.
  2. Chat over it, embedded as a widget, Slack bot, or CLI.
  3. Cite result.sources so users can click through to the real page.
from ragrails import RagRails

rag = RagRails()
rag.setup_url()

rag.ingest(
    urls="https://docs.example.com",
    ingestion={"mode": "full", "max_pages": 500},
    embedding={"provider": "voyage", "model": "voyage-3"},
    storage={"vector_db": "qdrant", "collection": "docs", "url": "http://localhost:6333"},
)

llm = rag.llm(provider="openai", model="gpt-4o-mini")
embedder = rag.embedder(provider="voyage", model="voyage-3", input_type="query")

result = rag.chat(
    "How do I rotate an API key?",
    llm=llm, embedder=embedder,
    vector_db="qdrant", collection="docs", url="http://localhost:6333",
    history=[],
)
print(result.answer)
for s in result.sources:
    print("·", s.get("title"))
Re-crawl on a schedule and use edit()/delete() to keep the index in sync as docs change.

​
Features used

Ingestion · Chat · Keeping your index current