Knowledge Base Setup

Create and configure knowledge bases for your agents

Attach file and URL sources so agents answer from your organisation’s documents.

from praisonaiagents import Agent, Knowledge

knowledge = Knowledge(sources=["docs/", "faq.pdf"])
agent = Agent(
    name="Support Agent",
    instructions="Answer using the knowledge base only.",
    knowledge=knowledge,
)
agent.start("What is our refund policy?")

The user asks a question; the agent retrieves relevant chunks and returns a grounded answer.

Add domain-specific knowledge to your agents using knowledge bases.

Quick Start

1. Simple Usage

   from praisonaiagents import Agent, Knowledge
   
   # Create knowledge base from files
   knowledge = Knowledge(
       sources=["docs/", "data.pdf", "faq.txt"]
   )
   
   # Create agent with knowledge
   agent = Agent(
       name="Support Agent",
       instructions="Answer questions using the knowledge base.",
       knowledge=knowledge
   )
   
   result = agent.start("What is your refund policy?")
   
2. With Configuration

   from praisonaiagents import Agent, Knowledge
   
   knowledge = Knowledge(
       sources=["docs/"],
       chunk_size=500,
       chunk_overlap=50,
   )
   
   agent = Agent(
       name="Support Agent",
       instructions="Answer using the knowledge base.",
       knowledge=knowledge,
   )
   agent.start("Summarise the onboarding guide.")
   

Knowledge Sources

From Files

knowledge = Knowledge(
    sources=[
        "documents/",      # Directory
        "manual.pdf",      # PDF
        "faq.md",          # Markdown
        "data.csv"         # CSV
    ]
)

From URLs

knowledge = Knowledge(
    sources=[
        "https://docs.example.com/api",
        "https://example.com/faq"
    ]
)

From Text

knowledge = Knowledge(
    sources=[],
    texts=[
        "Company policy: All refunds within 30 days.",
        "Support hours: 9 AM - 5 PM EST"
    ]
)

Configuration

knowledge = Knowledge(
    sources=["docs/"],
    chunk_size=500,           # Characters per chunk
    chunk_overlap=50,         # Overlap between chunks
    embedding_model="text-embedding-3-small",
    vector_store="chroma",    # or "faiss", "qdrant"
    top_k=5                   # Results to retrieve
)

Related

Chunking Strategies

Optimise chunking

Knowledge Module

Full API reference