Bedrock Knowledge Bases Advanced

Bedrock Knowledge Bases enable Retrieval-Augmented Generation (RAG) by connecting foundation models to your organization's data. Documents are automatically chunked, embedded, and stored in a vector database, allowing the model to retrieve relevant context when answering questions.

Knowledge Base Architecture

Ingestion:
  S3 Bucket (documents) → Chunking → Embedding Model → Vector Store

Query:
  User Question → Embedding → Vector Search → Relevant Chunks
  Relevant Chunks + Question → Foundation Model → Grounded Answer

Supported Vector Stores:
  - Amazon OpenSearch Serverless (recommended)
  - Amazon Aurora PostgreSQL (pgvector)
  - Pinecone
  - Redis Enterprise Cloud

Setting Up a Knowledge Base

1. Prepare data source

   Upload documents (PDF, TXT, HTML, CSV, DOCX) to an S3 bucket.

2. Create knowledge base

   Select embedding model (Titan Embeddings), configure chunking strategy, and choose vector store.

3. Sync data

   Trigger ingestion to process documents into vector embeddings. Re-sync when documents change.

4. Query

   Use the RetrieveAndGenerate API or connect to a Bedrock Agent.

Chunking Strategies

Querying the Knowledge Base

client = boto3.client('bedrock-agent-runtime')

response = client.retrieve_and_generate(
    input={'text': 'What is our refund policy?'},
    retrieveAndGenerateConfiguration={
        'type': 'KNOWLEDGE_BASE',
        'knowledgeBaseConfiguration': {
            'knowledgeBaseId': 'KB_ID',
            'modelArn': 'arn:aws:bedrock:us-east-1::foundation-model/anthropic.claude-3-5-sonnet-20241022-v2:0'
        }
    }
)
print(response['output']['text'])