RAG

RAG (Retrieval-Augmented Generation) grounds model responses in your own data. A thin, composable starting point under ai/ that wires a vector store into the generation flow.

What RAG does

Instead of asking the model to answer from memory, RAG retrieves relevant documents first, injects them into the prompt, and lets the model answer from that context — fewer hallucinations, and the answer is traceable to a source.

Embedding and indexing

# ai/rag.py
from bionicjs.rag import VectorStore

store = VectorStore()  # backed by your vector database

store.ingest("docs.txt", chunk_size=512)

Querying with context

// server/api/search.ts
import { ai } from "@bionicjs/core/server";

export default defineEventHandler(async (event) => {
  const { question } = await readBody(event);
  const context = await ai.rag.query(question, { topK: 4 });
  return {
    answer: await ai.messages.create({
      model: ai.model,
      messages: [{ role: "user", content: `Answer using:\n${context}\n\nQuestion: ${question}` }],
    }),
    sources: context.sources,
  };
});

Good to know

• RAG composes with prompts and agents: retrieved context becomes the prompt, and the agent decides how to use it.
• The vector store is configuration, not framework magic — point the plugin at the store you already use.