🏋️ AI Search + Agent Service: Fitness-Fun Example 🤸¶

Welcome to our AI Search + AI Agent tutorial, where we'll:

Create an Azure AI Search index with some fitness-oriented sample data
Demonstrate how to connect that index to an Agent via the AzureAISearchTool
Show how to query the Agent for health and fitness info in a fun scenario (with disclaimers!)

🏥 Health & Fitness Disclaimer¶

This notebook is for general demonstration and entertainment purposes, NOT a substitute for professional medical advice. Always seek the advice of certified health professionals.

Prerequisites¶

A Microsoft Azure subscription.
An Azure AI Search resource (formerly "Cognitive Search"), with admin API key or role-based access.

Python 3.8+, plus these libraries:

pip install azure-search-documents==11.4.0
pip install azure-ai-projects azure-identity
pip install opentelemetry-sdk azure-core-tracing-opentelemetry

Environment variables or a way to store your secrets:
- SEARCH_ENDPOINT - The endpoint URL of your Azure AI Search service
- SEARCH_API_KEY - The admin key for your Azure AI Search service
- PROJECT_CONNECTION_STRING - Your Azure AI Foundry project connection string (from the overview page)
- MODEL_DEPLOYMENT_NAME - The name of your deployed model in Azure AI Foundry

High-Level Flow¶

We'll do the following:

Create an AI Search index programmatically with sample fitness data.
Upload documents (fitness items) to the index.
Create an Agent that references our new index using AzureAISearchTool.
Run queries to see how it fetches from the index.

1. Create & Populate Azure AI Search Index¶

We'll create a minimal index called myfitnessindex, containing a few example items. Make sure to set your environment variables for SEARCH_ENDPOINT and SEARCH_API_KEY. We'll use the azure.search.documents.indexes classes to manage the index schema. We'll also upload some sample data.

In [ ]:

  Copied!     
 
import os
from azure.core.credentials import AzureKeyCredential
from azure.search.documents.indexes import SearchIndexClient
from azure.search.documents.indexes.models import SearchIndex, SimpleField, SearchFieldDataType
from azure.search.documents import SearchClient

search_endpoint = os.environ.get("SEARCH_ENDPOINT")
search_api_key = os.environ.get("SEARCH_API_KEY")

index_name = "myfitnessindex"

try:
    credential = AzureKeyCredential(search_api_key)
    index_client = SearchIndexClient(endpoint=search_endpoint, credential=credential)
    print("✅ SearchIndexClient created")
except Exception as e:
    print(f"❌ Error creating SearchIndexClient: {e}")
import os from azure.core.credentials import AzureKeyCredential from azure.search.documents.indexes import SearchIndexClient from azure.search.documents.indexes.models import SearchIndex, SimpleField, SearchFieldDataType from azure.search.documents import SearchClient search_endpoint = os.environ.get("SEARCH_ENDPOINT") search_api_key = os.environ.get("SEARCH_API_KEY") index_name = "myfitnessindex" try: credential = AzureKeyCredential(search_api_key) index_client = SearchIndexClient(endpoint=search_endpoint, credential=credential) print("✅ SearchIndexClient created") except Exception as e: print(f"❌ Error creating SearchIndexClient: {e}")

Define the index schema with a FitnessItemID key and a few fields to store product info.

In [ ]:

  Copied!     
 
def create_fitness_index():
    fields = [
        SimpleField(name="FitnessItemID", type=SearchFieldDataType.String, key=True),
        SimpleField(name="Name", type=SearchFieldDataType.String, filterable=True, searchable=True),
        SimpleField(name="Category", type=SearchFieldDataType.String, filterable=True, facetable=True, searchable=True),
        SimpleField(name="Price", type=SearchFieldDataType.Double, filterable=True, sortable=True, facetable=True),
        SimpleField(name="Description", type=SearchFieldDataType.String, searchable=True)
    ]

    index = SearchIndex(name=index_name, fields=fields)

    # If index already exists, we can delete first to start fresh (optional)
    if index_name in [x.name for x in index_client.list_indexes()]:
        index_client.delete_index(index_name)
        print(f"🗑️ Deleted existing index: {index_name}")

    # Create brand-new index
    created = index_client.create_index(index)
    print(f"🎉 Created index: {created.name}")

# Create the index
create_fitness_index()
def create_fitness_index(): fields = [ SimpleField(name="FitnessItemID", type=SearchFieldDataType.String, key=True), SimpleField(name="Name", type=SearchFieldDataType.String, filterable=True, searchable=True), SimpleField(name="Category", type=SearchFieldDataType.String, filterable=True, facetable=True, searchable=True), SimpleField(name="Price", type=SearchFieldDataType.Double, filterable=True, sortable=True, facetable=True), SimpleField(name="Description", type=SearchFieldDataType.String, searchable=True) ] index = SearchIndex(name=index_name, fields=fields) # If index already exists, we can delete first to start fresh (optional) if index_name in [x.name for x in index_client.list_indexes()]: index_client.delete_index(index_name) print(f"🗑️ Deleted existing index: {index_name}") # Create brand-new index created = index_client.create_index(index) print(f"🎉 Created index: {created.name}") # Create the index create_fitness_index()

Upload some sample documents to myfitnessindex. We'll add a few items for demonstration.

In [ ]:

  Copied!     
 
def upload_fitness_docs():
    # Construct a search client for data upload
    search_client = SearchClient(endpoint=search_endpoint,
                                 index_name=index_name,
                                 credential=AzureKeyCredential(search_api_key))

    sample_docs = [
        {
            "FitnessItemID": "1",
            "Name": "Adjustable Dumbbell",
            "Category": "Strength",
            "Price": 59.99,
            "Description": "A compact, adjustable weight for targeted muscle workouts."
        },
        {
            "FitnessItemID": "2",
            "Name": "Yoga Mat",
            "Category": "Flexibility",
            "Price": 25.0,
            "Description": "Non-slip mat designed for yoga, Pilates, and other exercises."
        },
        {
            "FitnessItemID": "3",
            "Name": "Treadmill",
            "Category": "Cardio",
            "Price": 499.0,
            "Description": "A sturdy treadmill with adjustable speed and incline settings."
        },
        {
            "FitnessItemID": "4",
            "Name": "Resistance Bands",
            "Category": "Strength",
            "Price": 15.0,
            "Description": "Set of colorful bands for light to moderate resistance workouts."
        }
    ]

    # Upload in one go
    result = search_client.upload_documents(documents=sample_docs)
    print(f"🚀 Upload result: {result}")

upload_fitness_docs()
def upload_fitness_docs(): # Construct a search client for data upload search_client = SearchClient(endpoint=search_endpoint, index_name=index_name, credential=AzureKeyCredential(search_api_key)) sample_docs = [ { "FitnessItemID": "1", "Name": "Adjustable Dumbbell", "Category": "Strength", "Price": 59.99, "Description": "A compact, adjustable weight for targeted muscle workouts." }, { "FitnessItemID": "2", "Name": "Yoga Mat", "Category": "Flexibility", "Price": 25.0, "Description": "Non-slip mat designed for yoga, Pilates, and other exercises." }, { "FitnessItemID": "3", "Name": "Treadmill", "Category": "Cardio", "Price": 499.0, "Description": "A sturdy treadmill with adjustable speed and incline settings." }, { "FitnessItemID": "4", "Name": "Resistance Bands", "Category": "Strength", "Price": 15.0, "Description": "Set of colorful bands for light to moderate resistance workouts." } ] # Upload in one go result = search_client.upload_documents(documents=sample_docs) print(f"🚀 Upload result: {result}") upload_fitness_docs()

Verify the documents via a basic query¶

Let's do a quick search for Strength items.

In [ ]:

  Copied!     
 
# Quick check
search_client = SearchClient(endpoint=search_endpoint,
                             index_name=index_name,
                             credential=AzureKeyCredential(search_api_key))

results = search_client.search(query="Strength", filter=None, top=10)
for doc in results:
    print(doc)
# Quick check search_client = SearchClient(endpoint=search_endpoint, index_name=index_name, credential=AzureKeyCredential(search_api_key)) results = search_client.search(query="Strength", filter=None, top=10) for doc in results: print(doc) 

2. Create Agent With AI Search Tool¶

We'll create a new agent and attach an AzureAISearchTool referencing myfitnessindex. In your environment, you need:

PROJECT_CONNECTION_STRING - from your AI Foundry project overview
MODEL_DEPLOYMENT_NAME - from the deployed model name

Let's initialize the AIProjectClient with DefaultAzureCredential.

In [ ]:

  Copied!     
 
import os
from azure.identity import DefaultAzureCredential
from azure.ai.projects import AIProjectClient
from azure.ai.projects.models import AzureAISearchTool, ConnectionType

try:
    project_client = AIProjectClient.from_connection_string(
        credential=DefaultAzureCredential(),
        conn_str=os.environ["PROJECT_CONNECTION_STRING"],
    )
    print("✅ Successfully initialized AIProjectClient")
except Exception as e:
    print(f"❌ Error initializing project client: {e}")
import os from azure.identity import DefaultAzureCredential from azure.ai.projects import AIProjectClient from azure.ai.projects.models import AzureAISearchTool, ConnectionType try: project_client = AIProjectClient.from_connection_string( credential=DefaultAzureCredential(), conn_str=os.environ["PROJECT_CONNECTION_STRING"], ) print("✅ Successfully initialized AIProjectClient") except Exception as e: print(f"❌ Error initializing project client: {e}")

Find or create the Azure AI Search connection within your Foundry project¶

If you've already created a connection for your Azure AI Search resource in the Foundry project, you can list them and pick the correct one. If not, you can create a new connection in the Portal or CLI usage for Foundry.

In [ ]:

  Copied!     
 
conn_id = None
all_connections = project_client.connections.list()
for c in all_connections:
    if c.connection_type == ConnectionType.AZURE_AI_SEARCH:
        conn_id = c.id
        print(f"Found existing Azure AI Search connection: {conn_id}")
        break

if not conn_id:
    print("❌ No Azure AI Search connection found in your project.\n",
          "Please create one or ask your admin to do so.")
conn_id = None all_connections = project_client.connections.list() for c in all_connections: if c.connection_type == ConnectionType.AZURE_AI_SEARCH: conn_id = c.id print(f"Found existing Azure AI Search connection: {conn_id}") break if not conn_id: print("❌ No Azure AI Search connection found in your project.\n", "Please create one or ask your admin to do so.")

Create the Agent with `AzureAISearchTool`¶

We'll attach the tool, specifying the index name we created.

In [ ]:

  Copied!     
 
# Replace with your model deployment name from environment
model_name = os.environ.get("MODEL_DEPLOYMENT_NAME", "my-model-deployment")

agent = None
if conn_id:
    # Initialize the search tool with your index
    ai_search_tool = AzureAISearchTool(
        index_connection_id=conn_id,
        index_name=index_name  # our myfitnessindex
    )

    # Create the agent
    with project_client:
        agent = project_client.agents.create_agent(
            model=model_name,
            name="fitness-agent-search",
            instructions="""
            You are a Fitness Shopping Assistant. You help users find items, but always disclaim not to provide medical advice.
            """,
            tools=ai_search_tool.definitions,
            tool_resources=ai_search_tool.resources,
            headers={"x-ms-enable-preview": "true"},
        )
        print(f"🎉 Created agent, ID: {agent.id}")
# Replace with your model deployment name from environment model_name = os.environ.get("MODEL_DEPLOYMENT_NAME", "my-model-deployment") agent = None if conn_id: # Initialize the search tool with your index ai_search_tool = AzureAISearchTool( index_connection_id=conn_id, index_name=index_name # our myfitnessindex ) # Create the agent with project_client: agent = project_client.agents.create_agent( model=model_name, name="fitness-agent-search", instructions=""" You are a Fitness Shopping Assistant. You help users find items, but always disclaim not to provide medical advice. """, tools=ai_search_tool.definitions, tool_resources=ai_search_tool.resources, headers={"x-ms-enable-preview": "true"}, ) print(f"🎉 Created agent, ID: {agent.id}")

3. Run a Conversation with the Agent¶

We'll open a new thread, post a question, and let the agent search the index for relevant items.

In [ ]:

  Copied!     
 
def run_agent_query(question: str):
    # Create a new thread
    thread = project_client.agents.create_thread()
    print(f"📝 Created thread, ID: {thread.id}")

    # Create a user message
    message = project_client.agents.create_message(
        thread_id=thread.id,
        role="user",
        content=question
    )
    print(f"💬 Created user message, ID: {message.id}")

    # Create and process agent run
    run = project_client.agents.create_and_process_run(
        thread_id=thread.id,
        assistant_id=agent.id
    )
    print(f"🤖 Agent run status: {run.status}")

    if run.last_error:
        print("⚠️ Run error:", run.last_error)

    # Retrieve all messages in the thread
    msg_list = project_client.agents.list_messages(thread_id=thread.id)
    # We'll print the assistant's last reply
    for m in reversed(msg_list.data):
        if m.role == "assistant" and m.content:
            print("\nAssistant says:")
            for c in m.content:
                if hasattr(c, "text"):
                    print(c.text.value)
            break

# Let's try some queries
if agent:
    run_agent_query("Which items are good for strength training?")
    run_agent_query("I need something for cardio under $300, any suggestions?")
def run_agent_query(question: str): # Create a new thread thread = project_client.agents.create_thread() print(f"📝 Created thread, ID: {thread.id}") # Create a user message message = project_client.agents.create_message( thread_id=thread.id, role="user", content=question ) print(f"💬 Created user message, ID: {message.id}") # Create and process agent run run = project_client.agents.create_and_process_run( thread_id=thread.id, assistant_id=agent.id ) print(f"🤖 Agent run status: {run.status}") if run.last_error: print("⚠️ Run error:", run.last_error) # Retrieve all messages in the thread msg_list = project_client.agents.list_messages(thread_id=thread.id) # We'll print the assistant's last reply for m in reversed(msg_list.data): if m.role == "assistant" and m.content: print("\nAssistant says:") for c in m.content: if hasattr(c, "text"): print(c.text.value) break # Let's try some queries if agent: run_agent_query("Which items are good for strength training?") run_agent_query("I need something for cardio under $300, any suggestions?")

4. Cleanup¶

We'll clean up the agent. (In production, you might want to keep it!)

In [ ]:

  Copied!     
 
if agent:
    project_client.agents.delete_agent(agent.id)
    print("🗑️ Deleted agent")

index_client.delete_index(index_name)
print(f"🗑️ Deleted index {index_name}")
if agent: project_client.agents.delete_agent(agent.id) print("🗑️ Deleted agent") index_client.delete_index(index_name) print(f"🗑️ Deleted index {index_name}")

🎉 Congrats!¶

You've successfully:

Created an Azure AI Search index programmatically.
Populated it with sample fitness data.
Created an Agent that queries the index using AzureAISearchTool.
Asked the agent for item recommendations.

Continue exploring how to integrate OpenTelemetry or the azure-ai-evaluation library for advanced tracing and evaluation capabilities. Have fun, and stay fit! 🏆

🏋️ AI Search + Agent Service: Fitness-Fun Example 🤸¶

🏥 Health & Fitness Disclaimer¶

Prerequisites¶

High-Level Flow¶

1. Create & Populate Azure AI Search Index¶

Verify the documents via a basic query¶

2. Create Agent With AI Search Tool¶

Find or create the Azure AI Search connection within your Foundry project¶

Create the Agent with AzureAISearchTool¶

3. Run a Conversation with the Agent¶

4. Cleanup¶

🎉 Congrats!¶

Create the Agent with `AzureAISearchTool`¶