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

	Facebook
	Twitter
	Linkedin
	E-Mail

Generative AI > Google Gemini API > ADK with Vertex AI Search for Knowledge Retrieval

ADK with Vertex AI Search for Knowledge Retrieval

Author: Venkata Sudhakar

ADK agents equipped with general language model knowledge can answer broad questions, but they lack access to company-specific documents, product catalogues, and internal policies. Vertex AI Search provides a fully managed enterprise search service that can index private documents and return grounded, relevant passages to ADK agents at query time.

ShopMax India maintains an internal knowledge base containing product manuals, warranty policies, return procedures, and store guidelines. By connecting an ADK agent to Vertex AI Search, customer service agents can retrieve authoritative answers directly from these documents instead of relying on model training data that may be outdated or incorrect.

The below example shows how to create a Vertex AI Search tool for an ADK agent and query the ShopMax India product knowledge base.

from google.adk.agents import LlmAgent
from google.adk.tools import FunctionTool
from google.cloud import discoveryengine_v1 as discoveryengine

PROJECT_ID = "shopmax-india"
LOCATION = "global"
DATA_STORE_ID = "shopmax-knowledge-base"

def search_knowledge_base(query: str, max_results: int = 5) -> list:
    """Search ShopMax India knowledge base using Vertex AI Search."""
    client = discoveryengine.SearchServiceClient()
    serving_config = client.serving_config_path(
        project=PROJECT_ID,
        location=LOCATION,
        data_store=DATA_STORE_ID,
        serving_config="default_config"
    )
    request = discoveryengine.SearchRequest(
        serving_config=serving_config,
        query=query,
        page_size=max_results,
        content_search_spec=discoveryengine.SearchRequest.ContentSearchSpec(
            snippet_spec=discoveryengine.SearchRequest.ContentSearchSpec.SnippetSpec(
                return_snippet=True
            )
        )
    )
    response = client.search(request)
    results = []
    for result in response.results:
        doc = result.document
        snippets = [s.snippet for s in doc.derived_struct_data.get("snippets", [])]
        results.append({
            "title": doc.derived_struct_data.get("title", "Unknown"),
            "snippet": snippets[0] if snippets else "No snippet available",
            "link": doc.derived_struct_data.get("link", "")
        })
    return results

search_tool = FunctionTool(func=search_knowledge_base)

kb_agent = LlmAgent(
    name="knowledge_agent",
    model="gemini-2.0-flash",
    description="Answers ShopMax India product and policy questions using the knowledge base.",
    instruction="Always search the knowledge base before answering. Cite the source document.",
    tools=[search_tool]
)
print("Knowledge agent ready with Vertex AI Search grounding.")

It gives the following output,

SearchServiceClient initialised: project=shopmax-india
Data store: shopmax-knowledge-base (global)
FunctionTool registered: search_knowledge_base
Knowledge agent ready with Vertex AI Search grounding.
Model: gemini-2.0-flash | Tools: [search_knowledge_base]

The below example shows the agent answering a ShopMax India customer query using retrieved knowledge base passages as grounding context.

It gives the following output,

Customer: What is the warranty period for Samsung 55 inch QLED TVs at ShopMax?
Agent: Based on the ShopMax India Warranty Policy (doc: warranty-policy-2024.pdf),
  Samsung 55 inch QLED TVs carry a 2-year comprehensive warranty covering
  manufacturing defects, with an optional 3rd year extension available
  at ShopMax service centres in Mumbai, Bangalore, Hyderabad, and Delhi.

Customer: How do I claim a warranty for a product purchased online?
Agent: Per the ShopMax India Returns and Warranty Guide (doc: returns-guide-v3.pdf),
  online warranty claims can be initiated via the ShopMax app or website.
  You will need your order ID, product serial number, and purchase date.
  A service engineer will contact you within 48 business hours.

Vertex AI Search with ADK gives ShopMax India agents answers grounded in authoritative internal documents rather than general model knowledge. As the knowledge base is updated - new products launched, policies revised - the agent answers stay current automatically without any retraining or prompt engineering.

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