Configuring settings to use the AI-powered capabilities in TrueSight Server Automation

This topic describes how to configure  BMC Helix Innovation Studio so that the TSSA Operations Analyzer agent can use a customer-provided Large Language Model (LLM) to process natural-language queries and provide operational insights. By completing this procedure, you enable the Operations Analyzer agent to interpret user queries, retrieve data from TrueSight Server Automation, and respond through supported interfaces such as Microsoft Teams.

Before you begin

Before you start configuring the settings in BMC Helix Innovation Studio, make sure that you have the required access and information to register a Large Language Model (LLM) and associate it with the Operations Analyzer agent.

Make sure that you have the following permissions:

• User must have the administrator role and permission to BMC Helix Innovation Studio.
• User must be able to access BMC HelixGPT Agent Studio.
• Connection details for a supported LLM, such as GPT-4.1 or Gemini 2.5.

Process overview

The following steps outline the sequence for configuring TSSA Operations Analyzer to use an LLM. Each task builds on the previous one and must be completed in sequence.

1. To obtain the LLM model details.
2. To configure the LLM model in BMC Helix Innovation Studio.
3. To verify the model ID in the Operations Analyzer agent.
4. To verify the model ID in the Operations Analyzer skill.
5. To validate the end-to-end configuration.

Task 1: To obtain the LLM model details

Before configuring BMC Helix Innovation Studio, obtain the required metadata for your LLM model from your LLM provider. 

• Model endpoint URL 
• API key 
• Model ID (for example, GPT-4.1 or Gemini 2.5) 

Task 2: To configure the LLM model in BMC Helix Innovation Studio

In this task, you register the LLM by creating or updating a model record. This makes the model available to agents and skills.

To add a new model

Use the following procedure for registering a new model for the first time.

1. Log in to BMC Helix Innovation Studio as an administrator. 
2. On the Workspace tab, from the Application launcher, select HelixGPT Agent Studio.
   
   
3. Select the Model record definition, and click Edit data. 
   
   
4. On the Data editor (Model), click New to add a model record. 
5. Provide the following values:
   • API Endpoint URL
   • API Key
   • Model ID 
6. Click Save. 

To edit an existing model

To modify credentials or identifiers for an existing model, perform the following steps:

1. Log in to BMC Helix Innovation Studio with administrator privileges. 
2. On the Workspace tab, from the Application launcher, select HelixGPT Agent Studio.
   
   
3. Open the Model record definition, and click Edit data. 
4. Select the existing model record and click Edit. 
5. Update the API Endpoint URL, API Key, or Model ID as required. 
6. Click Save. 

After you save the record, the model is registered and becomes available to agents and skills. 

Task 3: To verify the model ID in the TSSA Operations Analyzer agent

After registering the model, ensure that the Operations Analyzer agent references the correct Model ID. This step ensures that the agent uses the intended LLM for query processing.

To verify the agent configuration 

1. Open the TSSA Operations Analyzer Agent in BMC HelixGPT Agent Studio. 
2. Locate the Model ID field. 
3. Confirm that the value matches the model configured in Task 2. 
4. Update the Model ID if required and click Save. 

Task 4: To verify the model ID in the Operations Analyzer skill

The Operations Analyzer skill must use the same Model ID as the agent to ensure consistent behavior across conversational interactions.

To verify the skill configuration 

1. In BMC HelixGPT Agent Studio, open the TSSA Operations Analyzer Skill. 
2. Verify that the Model ID matches the value configured for the agent. 
3. Update the configuration and click Save if required. 

Task 5: To validate the end-to-end configuration

After completing all configurations, verify that the entire setup functions correctly. This ensures that Operations Analyzer can process user queries reliably.

Verify that:

• The LLM model has been successfully registered in BMC Helix Innovation Studio. 
• The same Model ID is configured in both the Operations Analyzer Agent and the Operations Analyzer Skill. 
• The model endpoint is reachable, and the provided API key is authorized. 
• All configuration changes are saved without errors. 

query_max_rows: Controls the maximum number of rows retrieved from the TSSA database for a query. The default value is 20. Increasing this value allows more data to be included in the response, but may increase processing time and LLM token usage. Retrieved rows are analyzed by the LLM to generate responses. Increasing this limit allows more data to be processed, but can increase token usage, so use higher values only if needed.

max_conversation_history_messages: Determines how many previous messages from the current conversation the agent uses to maintain context when responding to a new query. The default value is 10, which helps ensure relevant and coherent responses without excessive context carryover.

After validation, Operations Analyzer is ready to process user queries from supported interfaces such as Microsoft Teams.