feat: integrate Agent SDK into kernel, migrate to GlobalStandard model types, remove Greeting function, and update Orchestrator to handle greetings

.github/workflows/CAdeploy.yml

@@ -10,8 +10,8 @@ on:
     - cron: '0 6,18 * * *'  # Runs at 6:00 AM and 6:00 PM GMT
 
 env:
-  GPT_MIN_CAPACITY: 250
-  TEXT_EMBEDDING_MIN_CAPACITY: 40
+  GPT_MIN_CAPACITY: 200
+  TEXT_EMBEDDING_MIN_CAPACITY: 80
   BRANCH_NAME: ${{ github.head_ref || github.ref_name }}
 
 jobs:

docs/CustomizingAzdParameters.md

@@ -19,9 +19,10 @@ By default this template will use the environment name as the prefix to prevent
 | `AZURE_ENV_EMBEDDING_MODEL_CAPACITY` | integer | `80`              | Set the capacity for embedding model deployment.                                                     |
 | `AZURE_ENV_IMAGETAG`                  | string  | `latest`            | Set the image tag (allowed values: `latest`, `dev`, `hotfix`).                                             |
 | `AZURE_LOCATION`            | string  | `japaneast`         | Sets the Azure region for resource deployment.  |
-| `AZURE_ENV_LOG_ANALYTICS_WORKSPACE_ID`            | string  | `<Existing Workspace Id>`         | Reuses an existing Log Analytics Workspace instead of provisioning a new one.   |
+| `AZURE_ENV_LOG_ANALYTICS_WORKSPACE_ID`            | string  | Guide to get your [Existing Workspace ID](/docs/re-use-log-analytics.md)        | Reuses an existing Log Analytics Workspace instead of provisioning a new one.   |
 | `AZURE_EXISTING_AI_PROJECT_RESOURCE_ID`            | string  | `<Existing AI Foundry Project Resource Id>`         | Reuses an existing AI Foundry Project Resource Id instead of provisioning a new one.   |
 
+
 ## How to Set a Parameter
 To customize any of the above values, run the following command **before** `azd up`:
 

docs/DeploymentGuide.md

@@ -137,6 +137,14 @@ To adjust quota settings, follow these [steps](./AzureGPTQuotaSettings.md).
 
 </details>
 
+<details>
+
+  <summary><b>Reusing an Existing Log Analytics Workspace</b></summary>
+
+  Guide to get your [Existing Workspace ID](/docs/re-use-log-analytics.md)
+
+</details>
+
 ### Deploying with AZD
 
 Once you've opened the project in [Codespaces](#github-codespaces), [Dev Containers](#vs-code-dev-containers), or [locally](#local-environment), you can deploy it to Azure by following these steps:

docs/QuotaCheck.md

@@ -11,7 +11,7 @@ azd auth login
 
 ### 📌 Default Models & Capacities:
 ```
-gpt-4o-mini:30, text-embedding-ada-002:80
+gpt-4o-mini:200, text-embedding-ada-002:80
 ```
 ### 📌 Default Regions:
 ```
@@ -37,19 +37,19 @@ eastus, uksouth, eastus2, northcentralus, swedencentral, westus, westus2, southc
    ```
 ✔️ Check specific model(s) in default regions:
   ```
-  ./quota_check_params.sh --models gpt-4o-mini:30,text-embedding-ada-002:80
+  ./quota_check_params.sh --models gpt-4o-mini:200,text-embedding-ada-002:80
   ```
 ✔️ Check default models in specific region(s):
   ```
 ./quota_check_params.sh --regions eastus,westus
   ```
 ✔️ Passing Both models and regions:  
   ```
-  ./quota_check_params.sh --models gpt-4o-mini:30 --regions eastus,westus2
+  ./quota_check_params.sh --models gpt-4o-mini:200 --regions eastus,westus2
   ```
 ✔️ All parameters combined:
   ```
- ./quota_check_params.sh --models gpt-4o-mini:30,text-embedding-ada-002:80 --regions eastus,westus --verbose
+ ./quota_check_params.sh --models gpt-4o-mini:200,text-embedding-ada-002:80 --regions eastus,westus --verbose
   ```
 
 ### **Sample Output**

docs/re-use-log-analytics.md

@@ -0,0 +1,31 @@
+[← Back to *DEPLOYMENT* guide](/docs/DeploymentGuide.md#deployment-options--steps)
+
+# Reusing an Existing Log Analytics Workspace
+To configure your environment to use an existing Log Analytics Workspace, follow these steps:
+---
+### 1. Go to Azure Portal
+Go to https://portal.azure.com
+
+### 2. Search for Log Analytics
+In the search bar at the top, type "Log Analytics workspaces" and click on it and click on the workspace you want to use.
+
+![alt text](../docs/images/re_use_log/logAnalyticsList.png)
+
+### 3. Copy Resource ID
+In the Overview pane, Click on JSON View
+
+![alt text](../docs/images/re_use_log/logAnalytics.png)
+
+Copy Resource ID that is your Workspace ID
+
+![alt text](../docs/images/re_use_log/logAnalyticsJson.png)
+
+### 4. Set the Workspace ID in Your Environment
+Run the following command in your terminal
+```bash
+azd env set AZURE_ENV_LOG_ANALYTICS_WORKSPACE_ID '<Existing Log Analytics Workspace Id>'
+```
+Replace `<Existing Log Analytics Workspace Id>` with the value obtained from Step 3.
+
+### 5. Continue Deployment
+Proceed with the next steps in the [deployment guide](/docs/DeploymentGuide.md#deployment-options--steps).

infra/deploy_ai_foundry.bicep

@@ -37,7 +37,7 @@ var aiModelDeployments = [
     name: embeddingModel
     model: embeddingModel
     sku: {
-      name: 'Standard'
+      name: 'GlobalStandard'
       capacity: embeddingDeploymentCapacity
     }
     raiPolicyName: 'Microsoft.Default'

infra/main.bicep

@@ -63,7 +63,7 @@ param imageTag string = 'latest'
     type: 'location'
     usageName: [
       'OpenAI.GlobalStandard.gpt-4o-mini,200'
-      'OpenAI.Standard.text-embedding-ada-002,80'
+      'OpenAI.GlobalStandard.text-embedding-ada-002,80'
     ]
   }
 })

infra/main.json

@@ -5,7 +5,7 @@
     "_generator": {
       "name": "bicep",
       "version": "0.36.177.2456",
-      "templateHash": "1860841716622591379"
+      "templateHash": "2238194529646818649"
     }
   },
   "parameters": {
@@ -114,7 +114,7 @@
           "type": "location",
           "usageName": [
             "OpenAI.GlobalStandard.gpt-4o-mini,200",
-            "OpenAI.Standard.text-embedding-ada-002,80"
+            "OpenAI.GlobalStandard.text-embedding-ada-002,80"
           ]
         },
         "description": "Location for AI Foundry deployment. This is the location where the AI Foundry resources will be deployed."
@@ -744,7 +744,7 @@
             "_generator": {
               "name": "bicep",
               "version": "0.36.177.2456",
-              "templateHash": "15524709584492116446"
+              "templateHash": "1124249040831466979"
             }
           },
           "parameters": {
@@ -1038,7 +1038,7 @@
                 "name": "[parameters('embeddingModel')]",
                 "model": "[parameters('embeddingModel')]",
                 "sku": {
-                  "name": "Standard",
+                  "name": "GlobalStandard",
                   "capacity": "[parameters('embeddingDeploymentCapacity')]"
                 },
                 "raiPolicyName": "Microsoft.Default"

infra/scripts/checkquota.sh

@@ -33,7 +33,7 @@ echo "✅ Azure subscription set successfully."
 # Define models and their minimum required capacities
 declare -A MIN_CAPACITY=(
     ["OpenAI.GlobalStandard.gpt-4o-mini"]=$GPT_MIN_CAPACITY
-    ["OpenAI.Standard.text-embedding-ada-002"]=$TEXT_EMBEDDING_MIN_CAPACITY
+    ["OpenAI.GlobalStandard.text-embedding-ada-002"]=$TEXT_EMBEDDING_MIN_CAPACITY
 )
 
 VALID_REGION=""

infra/scripts/copy_kb_files.sh

@@ -123,46 +123,54 @@ echo "Uploading files to Azure Blob Storage"
 # Using az storage blob upload-batch to upload files with managed identity authentication, as the az storage fs directory upload command is not working with managed identity authentication.
 az storage blob upload-batch --account-name "$storageAccount" --destination data/"$extractedFolder1" --source $extractionPath1 --auth-mode login --pattern '*' --overwrite --output none
 if [ $? -ne 0 ]; then
-    retries=3
+    maxRetries=5
+    retries=$maxRetries
     sleepTime=10
-    echo "Error: Failed to upload files to Azure Blob Storage. Retrying upload...($((4 - retries)) of 3)"
+    attempt=1
     while [ $retries -gt 0 ]; do
+        echo "Error: Failed to upload files to Azure Blob Storage. Retrying upload...$attempt of $maxRetries in $sleepTime seconds"
         sleep $sleepTime
         az storage blob upload-batch --account-name "$storageAccount" --destination data/"$extractedFolder1" --source $extractionPath1 --auth-mode login --pattern '*' --overwrite --output none
         if [ $? -eq 0 ]; then
             echo "Files uploaded successfully to Azure Blob Storage."
             break
         else
             ((retries--))
-            echo "Retrying upload... ($((4 - retries)) of 3)"
+            ((attempt++))
             sleepTime=$((sleepTime * 2))
-            sleep $sleepTime
         fi
     done
-    exit 1
+    if [ $retries -eq 0 ]; then
+        echo "Error: Failed to upload files after all retry attempts."
+        exit 1
+    fi
 else
     echo "Files uploaded successfully to Azure Blob Storage."
 fi
 
 az storage blob upload-batch --account-name "$storageAccount" --destination data/"$extractedFolder2" --source $extractionPath2 --auth-mode login --pattern '*' --overwrite --output none
 if [ $? -ne 0 ]; then
-    retries=3
+    maxRetries=5
+    retries=$maxRetries
+    attempt=1
     sleepTime=10
-    echo "Error: Failed to upload files to Azure Blob Storage. Retrying upload...($((4 - retries)) of 3)"
     while [ $retries -gt 0 ]; do
+        echo "Error: Failed to upload files to Azure Blob Storage. Retrying upload...$attempt of $maxRetries in $sleepTime seconds"
         sleep $sleepTime
         az storage blob upload-batch --account-name "$storageAccount" --destination data/"$extractedFolder2" --source $extractionPath2 --auth-mode login --pattern '*' --overwrite --output none
         if [ $? -eq 0 ]; then
             echo "Files uploaded successfully to Azure Blob Storage."
             break
         else
             ((retries--))
-            echo "Retrying upload... ($((4 - retries)) of 3)"
+            ((attempt++))
             sleepTime=$((sleepTime * 2))
-            sleep $sleepTime
         fi
     done
-    exit 1
+    if [ $retries -eq 0 ]; then
+        echo "Error: Failed to upload files after all retry attempts."
+        exit 1
+    fi
 else
     echo "Files uploaded successfully to Azure Blob Storage."
 fi

infra/scripts/quota_check_params.sh

@@ -47,7 +47,7 @@ log_verbose() {
 }
 
 # Default Models and Capacities (Comma-separated in "model:capacity" format)
-DEFAULT_MODEL_CAPACITY="gpt-4o-mini:30,text-embedding-ada-002:80"
+DEFAULT_MODEL_CAPACITY="gpt-4o-mini:200,text-embedding-ada-002:80"
 
 # Convert the comma-separated string into an array
 IFS=',' read -r -a MODEL_CAPACITY_PAIRS <<< "$DEFAULT_MODEL_CAPACITY"
@@ -165,11 +165,7 @@ for REGION in "${REGIONS[@]}"; do
         FOUND=false
         INSUFFICIENT_QUOTA=false
 
-        if [ "$MODEL_NAME" = "text-embedding-ada-002" ]; then
-            MODEL_TYPES=("openai.standard.$MODEL_NAME")
-        else
-            MODEL_TYPES=("openai.standard.$MODEL_NAME" "openai.globalstandard.$MODEL_NAME")
-        fi
+        MODEL_TYPES=("openai.standard.$MODEL_NAME" "openai.globalstandard.$MODEL_NAME")
 
         for MODEL_TYPE in "${MODEL_TYPES[@]}"; do
             FOUND=false

src/App/backend/agents/agent_factory.py

@@ -41,7 +41,8 @@ async def get_wealth_advisor_agent(cls):
                 )
 
                 agent_name = "WealthAdvisor"
-                agent_instructions = "You are a helpful assistant to a Wealth Advisor."
+                agent_instructions = '''You are a helpful assistant to a Wealth Advisor.
+                If the question is unrelated to data but is conversational (e.g., greetings or follow-ups), respond appropriately using context, do not use external tools or perform any web searches for these conversational inputs.'''
 
                 agent_definition = await client.agents.create_agent(
                     model=ai_agent_settings.model_deployment_name,
@@ -105,3 +106,46 @@ async def delete_all_agent_instance(cls):
                 )
                 cls._search_agent["client"].close()
                 cls._search_agent = None
+
+    @classmethod
+    async def get_sql_agent(cls) -> dict:
+        """
+        Get or create a singleton SQLQueryGenerator AzureAIAgent instance.
+        This agent is used to generate T-SQL queries from natural language input.
+        """
+        async with cls._lock:
+            if not hasattr(cls, "_sql_agent") or cls._sql_agent is None:
+
+                agent_instructions = config.SQL_SYSTEM_PROMPT or """
+    You are an expert assistant in generating T-SQL queries based on user questions.
+    Always use the following schema:
+    1. Table: Clients (ClientId, Client, Email, Occupation, MaritalStatus, Dependents)
+    2. Table: InvestmentGoals (ClientId, InvestmentGoal)
+    3. Table: Assets (ClientId, AssetDate, Investment, ROI, Revenue, AssetType)
+    4. Table: ClientSummaries (ClientId, ClientSummary)
+    5. Table: InvestmentGoalsDetails (ClientId, InvestmentGoal, TargetAmount, Contribution)
+    6. Table: Retirement (ClientId, StatusDate, RetirementGoalProgress, EducationGoalProgress)
+    7. Table: ClientMeetings (ClientId, ConversationId, Title, StartTime, EndTime, Advisor, ClientEmail)
+
+    Rules:
+    - Always filter by ClientId = <provided>
+    - Do not use client name for filtering
+    - Assets table contains snapshots by date; do not sum values across dates
+    - Use StartTime for time-based filtering (meetings)
+    - Only return the raw T-SQL query. No explanations or comments.
+    """
+
+                project_client = AIProjectClient(
+                    endpoint=config.AI_PROJECT_ENDPOINT,
+                    credential=DefaultAzureCredentialSync(),
+                    api_version="2025-05-01",
+                )
+
+                agent = project_client.agents.create_agent(
+                    model=config.AZURE_OPENAI_MODEL,
+                    instructions=agent_instructions,
+                    name="SQLQueryGeneratorAgent",
+                )
+
+                cls._sql_agent = {"agent": agent, "client": project_client}
+        return cls._sql_agent