Skip to main content

​
The problem

Ask an LLM to write about your product and it invents features, prices, and policies. Ungrounded content is a liability.

​
How Ragrails solves it

Retrieve the relevant facts first, then pass them to your generation prompt as context.
  1. Retrieve the supporting chunks for the topic.
  2. Generate with those chunks as grounding, using your LLM and your prompt.
from ragrails import RagRails

rag = RagRails()
embedder = rag.embedder(provider="voyage", model="voyage-3", input_type="query")

hits = rag.retrieve(
    "enterprise plan features and limits",
    embedder=embedder,
    vector_db="qdrant", collection="docs", url="http://localhost:6333",
    top_k=8,
)

context = "\n\n".join(h.text for h in hits.items)
prompt = f"Write a product brief using ONLY these facts:\n\n{context}"
# → pass `prompt` to your LLM of choice
Or use chat() directly if you want Ragrails to handle generation with grounding and citations built in.

​
Features used

Retrieval · Reranking · Chat