RAG - Document Knowledge Base

Store, search, and retrieve documents to give AI access to your knowledge.

What You'll Learn

• Overview - What RAG is and how it works
• Collections - Create and manage document collections
• Uploading - Upload documents to your knowledge base
• Searching - Search documents with semantic queries
• Managing - List, view, and delete documents
• Examples - Integration examples

Overview

RAG (Retrieval-Augmented Generation) combines your documents with AI. It searches your knowledge base and provides relevant context to the AI.

How RAG Works

1. Upload - Documents are parsed, chunked, and stored as vectors
2. Search - Queries are converted to vectors and matched against documents
3. Retrieve - Most relevant chunks are returned with similarity scores
4. Generate - AI uses retrieved context to answer your question

Benefits

Feature	Without RAG	With RAG
Access to your data	No	Yes
Accurate answers	Limited	Yes
Source citations	No	Yes
Up-to-date knowledge	No	Yes

Use Cases

Use Case	Description
Documentation	Product docs, API references, user guides
Knowledge Base	Internal policies, procedures, FAQs
Customer Support	Support articles, troubleshooting guides
Research	Papers, reports, analysis documents
Legal	Contracts, regulations, case law

Quick Example

# Upload a document
curl -X POST http://localhost/api/v1/rag/upload \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "file=@user_guide.pdf" \
  -F "collection_name=documentation"

# Search documents
curl -X POST http://localhost/api/v1/rag/search \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "collection_name": "documentation",
    "query": "How do I reset my password?",
    "top_k": 3
  }'

Supported File Formats

Format	Extension	Notes
Text	.txt	Plain text files
Markdown	.md	Markdown documentation
PDF	.pdf	PDF documents
Word	.docx	Microsoft Word
JSON	.json	Structured data
HTML	.html	Web pages

Document Processing Pipeline

When you upload a document:

1. Parse - Extract text from file
2. Chunk - Split into smaller pieces (500-1000 tokens)
3. Embed - Convert chunks to vectors
4. Store - Save vectors in vector database

RAG in Action

import requests

# Upload document
with open("product_docs.pdf", "rb") as f:
    response = requests.post(
        "http://localhost/api/v1/rag/upload",
        headers={"Authorization": "Bearer YOUR_API_KEY"},
        files={"file": f},
        data={"collection_name": "docs"}
    )
print(f"Uploaded: {response.json()['document_id']}")

# Search documents
response = requests.post(
    "http://localhost/api/v1/rag/search",
    headers={"Authorization": "Bearer YOUR_API_KEY"},
    json={
        "collection_name": "docs",
        "query": "How to configure API keys?",
        "top_k": 3
    }
)

results = response.json()["results"]
for result in results:
    print(f"Score: {result['score']:.2f}")
    print(f"Content: {result['content'][:100]}...\n")

Next Steps

• Create Collections - Set up document collections
• Upload Documents - Add documents to your knowledge base
• Search Documents - Find relevant information
• Examples - See RAG in action