Update instructions to better make use AI Search or File Search. Updated documentation (#130)

* Update instructions to better make use AI Search or File Search

* Fix documentation

Author: howieleung

.github/workflows/azure-dev.yml

@@ -37,7 +37,7 @@ jobs:
       AZURE_LOG_ANALYTICS_WORKSPACE_NAME: ${{ vars.AZURE_LOG_ANALYTICS_WORKSPACE_NAME }}
       USE_CONTAINER_REGISTRY: ${{ vars.USE_CONTAINER_REGISTRY }}
       USE_APPLICATION_INSIGHTS: ${{ vars.USE_APPLICATION_INSIGHTS }}
-      USE_SEARCH_SERVICE: ${{ vars.USE_SEARCH_SERVICE }}
+      USE_AZURE_AI_SEARCH_SERVICE: ${{ vars.USE_AZURE_AI_SEARCH_SERVICE }}
       AZURE_AI_AGENT_NAME: ${{ vars.AZURE_AI_AGENT_NAME }}
       AZURE_EXISTING_AGENT_ID: ${{ vars.AZURE_EXISTING_AGENT_ID }}
       AZURE_AI_AGENT_DEPLOYMENT_NAME: ${{ vars.AZURE_AI_AGENT_DEPLOYMENT_NAME }}

.github/workflows/template-validation.yml

@@ -33,7 +33,7 @@ jobs:
           AZURE_STORAGE_ACCOUNT_NAME: ${{ vars.AZURE_STORAGE_ACCOUNT_NAME }}
           AZURE_LOG_ANALYTICS_WORKSPACE_NAME: ${{ vars.AZURE_LOG_ANALYTICS_WORKSPACE_NAME }}
           USE_APPLICATION_INSIGHTS: ${{ vars.USE_APPLICATION_INSIGHTS }}
-          USE_SEARCH_SERVICE: ${{ vars.USE_SEARCH_SERVICE }}
+          USE_AZURE_AI_SEARCH_SERVICE: ${{ vars.USE_AZURE_AI_SEARCH_SERVICE }}
           AZURE_AI_AGENT_NAME: ${{ vars.AZURE_AI_AGENT_NAME }}
           AZURE_EXISTING_AGENT_ID: ${{ vars.AZURE_EXISTING_AGENT_ID }}
           AZURE_AI_AGENT_DEPLOYMENT_NAME: ${{ vars.AZURE_AI_AGENT_DEPLOYMENT_NAME }}

README.md

@@ -108,9 +108,9 @@ When you start a deployment, most parameters will have default values. You can c
 | **Azure Region** | Select a region with quota which supports your selected model. |   |
 | **Model** | Choose from the [list of models supported by Azure AI Agent Service](https://learn.microsoft.com/azure/ai-services/agents/concepts/model-region-support) for your selected region. | gpt-4o-mini |  
 | **Model Format** | Choose from OpenAI or Microsoft, depending on your model. | OpenAI |  
-| **Model Deployment Capacity** | Configure capacity for your model. Recommended value is 100k. | 30k |
+| **Model Deployment Capacity** | Configure capacity for your model. | 80k |
 | **Embedding Model** | Choose from text-embedding-3-large, text-embedding-3-small, and text-embedding-ada-002. This may only be deployed if Azure AI Search is enabled. |  text-embedding-3-small |
-| **Embedding Model Capacity** | Configure capacity for your embedding model. |  30k |
+| **Embedding Model Capacity** | Configure capacity for your embedding model. |  50k |
 | **Knowledge Retrieval** | Choose from OpenAI's file search or including Azure AI Search Index. |  OpenAI's file search |
 
 For a detailed description of customizable fields and instructions, view the [deployment customization guide](docs/deploy_customization.md).

azure.yaml

@@ -55,7 +55,7 @@ pipeline:
     - AZURE_LOG_ANALYTICS_WORKSPACE_NAME
     - USE_CONTAINER_REGISTRY
     - USE_APPLICATION_INSIGHTS
-    - USE_SEARCH_SERVICE
+    - USE_AZURE_AI_SEARCH_SERVICE
     - AZURE_AI_AGENT_NAME
     - AZURE_AI_AGENT_ID
     - AZURE_AI_AGENT_DEPLOYMENT_NAME

docs/deploy_customization.md

@@ -3,17 +3,27 @@
 
 This document describes how to customize the deployment of the Agents Chat with Azure AI Foundry. Once you follow the steps here, you can run `azd up` as described in the [Deploying](./../README.md#deploying-steps) steps.
 
-* [Disabling resources](#disabling-resources)
+* [Use existing resources](#use-existing-resources)
+* [Enabling and disabling resources provision](#enabling-and-disabling-resources-provision)
 * [Customizing resource names](#customizing-resource-names)
 * [Customizing model deployments](#customizing-model-deployments)
 
-## Disabling resources
+## Use existing resources
+Be default, this template provisions a new resource group along with other resources.   If you already have provisioned Azure AI Foundry and Azure AI Foundry Project, you might reuse these resources by setting:
 
-Disabling a resource will stop that resource from being created and deployed to your Azure Project. 
+```shell
+azd env set AZURE_EXISTING_AIPROJECT_RESOURCE_ID "https://<your-ai-services-account-name>.services.ai.azure.com/api/projects/<your-project-name>"
+```
+
+Notices that Application Insight and AI Search will not be created in this scenario.
+
+
+## Enabling and disabling resources provision
+
+By default, provisioning Application Insights is enabled, and AI Search is disabled.  The default setting can be changed by:
 
-* To disable AI Search, run `azd env set USE_SEARCH_SERVICE false`
+* To enable AI Search, run `azd env set USE_AZURE_AI_SEARCH_SERVICE true`
 * To disable Application Insights, run `azd env set USE_APPLICATION_INSIGHTS false`
-* To disable Container Registry, run `azd env set USE_CONTAINER_REGISTRY false`
 
 Once you disable these resources, they will not be deployed when you run `azd up`.
 
@@ -22,9 +32,8 @@ Once you disable these resources, they will not be deployed when you run `azd up
 By default, this template will use a naming convention with unique strings to prevent naming collisions within Azure.
 To override default naming conventions, the following keys can be set:
 
-* `AZURE_EXISTING_AIPROJECT_RESOURCE_ID` - An existing AI Project Resource ID be used.   If specified, resources for AI Foundry Project and Azure AI service will not be created.
-* `AZURE_AIPROJECT_NAME` - The name of the AI Foundry Project
-* `AZURE_AISERVICES_NAME` - The name of the Azure AI service
+* `AZURE_AIPROJECT_NAME` - The name of the Azure AI Foundry project
+* `AZURE_AISERVICES_NAME` - The name of the Azure AI Foundry
 * `AZURE_STORAGE_ACCOUNT_NAME` - The name of the Storage Account
 * `AZURE_APPLICATION_INSIGHTS_NAME` - The name of the Application Insights instance
 * `AZURE_LOG_ANALYTICS_WORKSPACE_NAME` - The name of the Log Analytics workspace used by Application Insights

src/gunicorn.conf.py

@@ -18,6 +18,7 @@
     AzureAISearchTool,
     FilePurpose,
     FileSearchTool,
+    Tool,
 )
 from azure.ai.projects.models import ConnectionType
 from azure.identity.aio import DefaultAzureCredential
@@ -120,9 +121,9 @@ def _get_file_path(file_name: str) -> str:
                      file_name))
 
 
-async def get_available_toolset(
+async def get_available_tool(
         project_client: AIProjectClient,
-        creds: AsyncTokenCredential) -> AsyncToolSet:
+        creds: AsyncTokenCredential) -> Tool:
     """
     Get the toolset and tool definition for the agent.
 
@@ -145,11 +146,9 @@ async def get_available_toolset(
     if conn_id:
         await create_index_maybe(project_client, creds)
 
-        ai_search = AzureAISearchTool(
+        return AzureAISearchTool(
             index_connection_id=conn_id,
             index_name=os.environ.get('AZURE_AI_SEARCH_INDEX_NAME'))
-
-        toolset.add(ai_search)
     else:
         logger.info(
             "agent: index was not initialized, falling back to file search.")
@@ -169,21 +168,22 @@ async def get_available_toolset(
         )
         logger.info("agent: file store and vector store success")
 
-        file_search_tool = FileSearchTool(vector_store_ids=[vector_store.id])
-        toolset.add(file_search_tool)
-
-    return toolset
+        return FileSearchTool(vector_store_ids=[vector_store.id])
 
 
 async def create_agent(ai_client: AIProjectClient,
                        creds: AsyncTokenCredential) -> Agent:
     logger.info("Creating new agent with resources")
-    toolset = await get_available_toolset(ai_client, creds)
-
+    tool = await get_available_tool(ai_client, creds)
+    toolset = AsyncToolSet()
+    toolset.add(tool)
+    
+    instructions = "Use AI Search always. Avoid to use base knowledge." if isinstance(tool, AzureAISearchTool) else "Use File Search always.  Avoid to use base knowledge."
+    
     agent = await ai_client.agents.create_agent(
         model=os.environ["AZURE_AI_AGENT_DEPLOYMENT_NAME"],
         name=os.environ["AZURE_AI_AGENT_NAME"],
-        instructions="You are helpful assistant",
+        instructions=instructions,
         toolset=toolset
     )
     return agent