Skip to content

RAG (Retrieval-Augmented Generation)

IdeaWeaver provides comprehensive RAG capabilities for building knowledge-based AI systems. The RAG system supports both traditional and agentic approaches, with multiple vector store options and advanced evaluation features.

Key Features

  • Multiple Vector Stores: Support for Chroma, Qdrant (local and cloud)
  • Advanced Embeddings: Sentence Transformers, OpenAI, and custom models
  • Document Processing: PDF, DOCX, TXT, Markdown with smart chunking
  • Evaluation: RAGAS framework for comprehensive assessment
  • Agentic RAG: Multi-step reasoning with tool integration

Qdrant Integration

IdeaWeaver provides seamless integration with Qdrant, a high-performance vector database. You can setup:

  1. Qdrant Cloud: For production deployments with managed service

Using Qdrant Local

Using Qdrant Cloud

# Set Qdrant Cloud credentials as environment variables
export QDRANT_URL="your-cloud-url"
export QDRANT_API_KEY="your-api-key"

ideaweaver rag create-kb --name cloud-kb \
    --vector-store qdrant_cloud \
    --qdrant-url "$QDRANT_URL" \
    --qdrant-api-key "$QDRANT_API_KEY" \
    --description "Cloud Qdrant knowledge base"

Note: For security reasons, never commit API keys or sensitive URLs to version control. Use environment variables or secure secret management systems.

Qdrant Features

  • High-performance vector search
  • Scalable architecture
  • Advanced filtering capabilities
  • Payload support for metadata
  • Real-time updates

RAG Architecture

RAG Architecture

Getting Started

  1. Create a knowledge base: 

    ideaweaver rag create-kb --name my-kb --description "My first knowledge base"

  2. Ingest documents: 

    ideaweaver rag ingest --kb my-kb --source ./docs --file-types md,pdf

  3. Query the knowledge base: 

    ideaweaver rag query --kb my-kb -q "What are the main features?"

  4. Evaluate the RAG system: 

    ideaweaver rag evaluate --kb my-kb

Next Steps