	In Browser
	StumbleUpon
	del.icio.us
	Google
	Google Buzz
	reddit
	LinkedIn

	Facebook
	Twitter
	Linkedin
	E-Mail

Generative AI > Google Gemini API > ADK Agents with Knowledge Graphs

ADK Agents with Knowledge Graphs

Author: Venkata Sudhakar

A knowledge graph stores entities and their relationships - products, brands, categories, compatibility links. For ShopMax India, this means an agent can answer questions like 'what cases fit my OnePlus 12?' or 'which earphones work with the Samsung S24?' without those answers being hard-coded. The agent queries the graph as a tool, traversing relationships at runtime.

The example below models the ShopMax India catalogue as a simple in-memory graph using a dictionary of nodes and edges. In production this would be Cloud Spanner or Neo4j. The agent uses two tools - one to look up product nodes, another to traverse compatibility edges.

from google.adk.agents import LlmAgent
from google.adk.tools import FunctionTool
from google.adk.sessions import InMemorySessionService
from google.adk.runners import Runner
from google.genai.types import Content, Part

# ShopMax India knowledge graph
NODES = {
    "P001": {"name": "OnePlus 12",          "type": "smartphone", "brand": "OnePlus",  "price_rs": 64999},
    "P002": {"name": "Samsung Galaxy S24",   "type": "smartphone", "brand": "Samsung",  "price_rs": 74999},
    "P003": {"name": "OnePlus Buds Pro 2",   "type": "earphones",  "brand": "OnePlus",  "price_rs": 9999},
    "P004": {"name": "Samsung Galaxy Buds2", "type": "earphones",  "brand": "Samsung",  "price_rs": 8999},
    "P005": {"name": "OnePlus 12 Silicone Case","type": "case",    "brand": "OnePlus",  "price_rs": 999},
    "P006": {"name": "Samsung S24 Clear Case","type": "case",      "brand": "Samsung",  "price_rs": 1199},
}

# Directed edges: product -> compatible accessories
EDGES = {
    "P001": ["P003", "P005"],  # OnePlus 12 -> Buds Pro 2, Silicone Case
    "P002": ["P004", "P006"],  # Galaxy S24 -> Galaxy Buds2, S24 Case
}

def lookup_product(product_name: str) -> dict:
    """Find a ShopMax India product node by name or partial name."""
    for pid, node in NODES.items():
        if product_name.lower() in node["name"].lower():
            return {"id": pid, **node}
    return {"error": "Product not found"}

def get_compatible_accessories(product_id: str) -> dict:
    """Get accessories compatible with a ShopMax India product by product ID."""
    acc_ids = EDGES.get(product_id, [])
    accessories = [{"id": aid, **NODES[aid]} for aid in acc_ids if aid in NODES]
    return {"product_id": product_id, "compatible_accessories": accessories}

agent = LlmAgent(
    name="shopmax_kg_agent",
    model="gemini-2.0-flash",
    tools=[FunctionTool(lookup_product), FunctionTool(get_compatible_accessories)],
    instruction=(
        "You are a ShopMax India product advisor. "
        "Use the knowledge graph tools to find products and their compatible accessories. "
        "Always look up the product first to get its ID, then fetch compatible accessories."
    )
)

session_service = InMemorySessionService()
runner = Runner(agent=agent, session_service=session_service, app_name="shopmax")

def ask(user_id: str, query: str):
    session = session_service.create_session(app_name="shopmax", user_id=user_id)
    events = list(runner.run(
        user_id=user_id,
        session_id=session.id,
        new_message=Content(parts=[Part(text=query)])
    ))
    for event in events:
        if hasattr(event, "content") and event.content:
            for part in event.content.parts:
                if hasattr(part, "text") and part.text:
                    print(f"Answer: {part.text.strip()[:250]}")

ask("u1", "What accessories are compatible with the OnePlus 12?")
print()
ask("u2", "I have a Samsung Galaxy S24 - what earphones should I buy?")

It gives the following output,

Answer: For the OnePlus 12 (Rs 64,999), ShopMax India has two compatible accessories:
1. OnePlus Buds Pro 2 - Rs 9,999 (wireless earphones with active noise cancellation)
2. OnePlus 12 Silicone Case - Rs 999 (official protective case)
Both are designed to work seamlessly with your OnePlus 12.

Answer: For your Samsung Galaxy S24, the best compatible earphones are the
Samsung Galaxy Buds2 at Rs 8,999. They pair automatically with your Galaxy S24
via Bluetooth and integrate with Samsung Sound settings for a seamless experience.

The agent traversed the knowledge graph in two steps - first calling lookup_product to get the product ID, then calling get_compatible_accessories to traverse the edges. This graph traversal pattern scales to hundreds of products and thousands of compatibility relationships without any hard-coding in the agent instruction.

For ShopMax India production, store the knowledge graph in Cloud Spanner using a node table and an edge table. The edge table has columns: source_product_id, target_product_id, relationship_type (compatible, replaces, bundles_with). The agent tools run parameterised Spanner queries - no escaped quotes needed. Refresh the graph nightly from the product catalogue database so new products and compatibility data are always current.

Send your comments, suggestions or queries regarding this site to [email protected].