Ask HelixGPT

Ask HelixGPT is a virtual chat agent available in the BMC Helix AIOps console for investigating ML-based situations. It provides a set of predefined questions that operators or site reliability engineers (SREs) can use to get information about an ML-based situation. The predefined questions help operators and SREs understand the overall impact of the situation, the team that can solve the issue, and restore the health of the impacted service.

The answers to these questions are generated by evaluating information from the incidents created for similar situations in BMC Helix ITSM, analyzing timestamps and patterns of similar situations that have occurred in the past, analyzing the service health score of the impacted service of the situation, and the change requests associated with the situation.

By using Ask HelixGPT operators or SREs can improve operational efficiency, derive insights from all connected sources, and reduce manual errors by implementing automation to resolve problems faster.

To use Ask HelixGPT, make sure that the required product licenses are available. For more information, see BMC Helix AIOps service.

Ask HelixGPT capabilities

Ask HelixGPT provides the following capabilities:

Predefined questions about the impact, team that can solve the issue, and whether the issue is a recurring one.
Generates answers by evaluating events, change requests, incidents, similar situations, and other data related to a situation.

To learn more about configuring Ask HelixGPT, see To set up Ask HelixGPT.

Scenario

Scenario: Ask predefined questions to investigate a situation faster

Susan is an SRE at Apex Global, responsible for ensuring the availability of the company’s customer-facing digital services. On a busy Monday morning, she notices that the Online Order Processing service is critically impacted. When she clicks on the service, she realizes that a situation has been created for the service, with dozens of events correlating to high CPU utilization, increased response times, and intermittent transaction failures, cutting through event noise. She opens the situation details page and clicks the Ask HelixGPT chat.

Instead of manually combing through metrics, logs, and dashboards, she asks the following question, "Impact?". The response includes a clear summary, confirming that the issue is affecting the checkout flow for customers across multiple regions and poses a risk of abandoned transactions if not resolved quickly. This gives Susan an immediate business context and helps her prioritize the incident.

Next, Susan asks, “Recurring?” The autonomous agents connected to Ask HelixGPT analyze historical data and reveal that a similar pattern occurred during peak traffic hours, indicating a known bottleneck in the application tier. This insight saves Susan valuable time by eliminating guesswork and narrowing the scope of investigation.

To avoid unnecessary escalations, Susan asks another predefined question, “Team to solve?”. Based on service topology, past resolutions, and ownership data, the Application Services team is named as the most relevant resolver group. With this information, Susan confidently routes the situation to the right team on the first attempt, no manual triage, no back-and-forth.

With clear impact analysis, historical context, and the correct ownership identified, the Application Services team quickly validates that a recent configuration change caused excessive resource consumption during peak load. The team rolls back the change and applies an optimized configuration. The situation is automatically resolved and closed, the service health returns to normal, and the incident is resolved using a single, consolidated incident record, minimizing operational noise.

Agent type, skills, and prompts

Ask HelixGPT supports the following agent types, skills, prompts, and supported models.

Agent type:
AIOps Chat Agent: Provides predefined questions that help identify the impact, re-occurance, team that might resolve the issue, and if any change requests are associated with the situation.
Out-of-the-box skill: No
Out-of-the-box prompts: No

User roles and permissions

Make sure that the following minimum roles are assigned to configure and set up Ask HelixGPT:

Product	Role	Description	Reference
BMC HelixGPT	HelixGPT Admin	Configure the model, update agent settings in BMC Helix Agent Studio.	Roles and permissions
BMC Helix AIOps	Tenant administrator	Obtain the fine-tuned BMC Helix AIOps model, deploy it in the cloud, and configure other settings in BMC HelixGPT.	Roles and permissions
BMC Helix AIOps	Operator or site reliability engineer (SRE)	Monitor and investigate situations in BMC Helix AIOps.	Roles and permissions

Process overview

The following flowchart explains the tasks required to configure and Ask HelixGPT:

Process overview_configuring AI agents for AIOps_26102.png

Process to set up Ask HelixGPT

The following table lists the tasks to set up and use Ask HelixGPT:

	Product	Task	Reference
1	BMC Helix AIOps	Obtain the fine-tuned BMC Helix AIOps model and deploy it in the cloud.	Configuring settings to use the AI-powered capabilities in BMC Helix AIOps
2	BMC HelixGPT	Update model settings and verify whether the same model ID is available for the agent	Configuring settings to use the AI-powered capabilities in BMC Helix AIOps

Ask HelixGPT use case

The following table describes the tasks that you can perform by using Ask HelixGPT:

Task	Description	Reference
Ask predefined questions to investigate an ML-based situation	Ask the following questions to investigate an ML-based situation faster, determine its impact, and identify the stakeholders that can help resolve it: Reoccuring? Change Windows? Impact Team to solve?	Investigating ML-based independent situations Investigating ML-based situations that impact multiple services