09-vertex-ai-search

09. Vertex AI Search

Enterprise-grade search integration with ADK - Build agents that search across websites, documents, and structured data.

Blog Post: https://arjunprabhulal.com/adk-builtin-tools-vertex-search/

Table of Contents

1. What is Vertex AI Search?
2. RAG vs Search: When to Use Which
3. Prerequisites
4. Google Cloud Setup
5. Create a Data Store
6. Setup Steps
7. Running the Agent
8. Project Structure
9. Next Steps

What is Vertex AI Search?

Vertex AI Search (formerly Enterprise Search) provides Google-quality search for your data:

┌─────────────────────────────────────────────────────────────┐
│                    DATA SOURCES                              │
│                                                              │
│   ┌──────────┐  ┌──────────┐  ┌──────────┐  ┌──────────┐   │
│   │ Websites │  │   PDFs   │  │ BigQuery │  │Cloud     │   │
│   │          │  │          │  │  Tables  │  │Storage   │   │
│   └──────────┘  └──────────┘  └──────────┘  └──────────┘   │
└─────────────────────────────────────────────────────────────┘
                            │
                            ▼
┌─────────────────────────────────────────────────────────────┐
│                    VERTEX AI SEARCH                          │
│                                                              │
│   • Automatic indexing and ranking                           │
│   • Semantic understanding                                   │
│   • Multi-modal search (text, images)                        │
│   • Extractive answers with citations                        │
└─────────────────────────────────────────────────────────────┘
                            │
                            ▼
┌─────────────────────────────────────────────────────────────┐
│                    ADK AGENT                                 │
│                                                              │
│   Uses VertexAiSearchTool to query and synthesize results    │
└─────────────────────────────────────────────────────────────┘

RAG vs Search: When to Use Which

Feature	Vertex AI RAG	Vertex AI Search
Best For	Q&A over documents	Enterprise search portals
Data Size	Smaller corpora	Large-scale data
Updates	Manual re-indexing	Automatic crawling
Sources	Uploaded files	Websites, GCS, BigQuery
Output	Generated answers	Search results + snippets

Use RAG when: You need precise answers from a curated document set.

Use Search when: You need to search across websites, large document collections, or structured data.

Prerequisites

• Python 3.10+
• Google Cloud Project with billing enabled
• gcloud CLI installed and authenticated
• Data to index (website URL, GCS bucket, or documents)

Google Cloud Setup

1. Enable Required APIs

gcloud services enable aiplatform.googleapis.com
gcloud services enable discoveryengine.googleapis.com

2. Grant IAM Roles

# Replace with your project ID and email
PROJECT_ID="your-project-id"
USER_EMAIL="your-email@example.com"

# Vertex AI User
gcloud projects add-iam-policy-binding $PROJECT_ID \
  --member="user:$USER_EMAIL" \
  --role="roles/aiplatform.user"

# Discovery Engine Viewer (or Editor to create data stores)
gcloud projects add-iam-policy-binding $PROJECT_ID \
  --member="user:$USER_EMAIL" \
  --role="roles/discoveryengine.viewer"

3. Authenticate

gcloud auth application-default login

Create a Data Store

Using Agent Builder Console

1. Go to Agent Builder Console
2. Click Create App → Select Search
3. Choose your data source type:
   • Website: Enter URLs to crawl
   • Cloud Storage: Select GCS bucket
   • BigQuery: Choose dataset
   • Unstructured Documents: Upload files
4. Configure indexing settings
5. Note the Data Store ID from the console

Data Store ID Format

projects/PROJECT_ID/locations/global/collections/default_collection/dataStores/DATASTORE_ID

Example:

projects/my-project/locations/global/collections/default_collection/dataStores/my-docs-store_1234567890

Setup Steps

1. Navigate to this module:

cd 09-vertex-ai-search

2. Create and activate a virtual environment:

python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

3. Install dependencies:

pip install -r ../requirements.txt

4. Configure environment variables in vertex_search_agent/.env:

GOOGLE_GENAI_USE_VERTEXAI=1
GOOGLE_CLOUD_PROJECT=your-project-id
GOOGLE_CLOUD_LOCATION=us-central1
DATA_STORE_ID=projects/your-project-id/locations/global/collections/default_collection/dataStores/your-datastore-id

Replace your-project-id and your-datastore-id with your actual values.

Running the Agent

Using ADK Web

adk web

Open http://127.0.0.1:8000 and select vertex_search_agent.

Test Queries:

• "Search for [topic in your data store]"
• "Find documents about [subject]"
• "What information do you have on [query]?"

Using ADK CLI

adk run vertex_search_agent

Project Structure

09-vertex-ai-search/
├── README.md
└── vertex_search_agent/
    ├── __init__.py
    ├── agent.py          # Search agent with VertexAiSearchTool
    └── .env              # Vertex AI configuration

Key Features

Feature	Description
Semantic Search	Understands meaning, not just keywords
Extractive Answers	Highlights exact passages that answer queries
Citations	Links back to source documents
Multi-turn	Supports follow-up questions
Filters	Filter by metadata, date, etc.

Next Steps

Continue to 10. Custom Function Tools