AI Agents · Lesson

Building a Knowledge-Augmented Agent

End-to-end: entity linking → graph query → answer synthesis.

What Is a Knowledge-Augmented Agent?

A knowledge-augmented agent enriches its answers using a knowledge base. When a question arrives, the agent extracts entities, looks them up in a knowledge graph, finds relevant documents via vector search, and provides all context to the LLM for a rich, grounded answer.

The Full Retrieval Pipeline

The agent pipeline: 1 receive question → 2 extract entities → 3 graph lookup for entity context → 4 vector search for relevant documents → 5 combine all context → 6 LLM generates answer.

from dataclasses import dataclass, field
from typing import List, Dict, Any

@dataclass
class RetrievalContext:
    question: str
    entities: List[str] = field(default_factory=list)
    graph_context: Dict[str, Any] = field(default_factory=dict)
    vector_documents: List[Dict] = field(default_factory=list)
    combined_context: str = ''
    answer: str = ''
    sources_used: List[str] = field(default_factory=list)

# The agent will populate this object as it works through the pipeline
ctx = RetrievalContext(question='What AI projects is Sam Altman known for?')
print('RetrievalContext created:', ctx.question)

All lessons in this course

← Back to AI Agents