Quick start into prompts

Last updated
Save as PDF

Quick start into prompting, including examples to copy & paste.

Understanding Large Language Models

Working with Configurable AI and prompts requires some understanding of Large Language Models to optimize the result of your use case. Here are a few key instructions:

Be very concrete in your instructions and avoid ambiguity.
Keep sentences short to make sure your intention is grasped by the large language models.
Provide context and role with sufficient background from your configuration - think about how the message needs to be formed in order to be useful for support agents.
- Let's break down a shorter example for Generate prompt here:
  - Provide a role and context - for example: “Act as an IT support service desk agent handling issues related to workstations and printers."
  - Introduce a background - for example: “You might be provided with the ongoing email conversation and with the data about the support ticket the agent is working on."
  - Add general instructions - for example: “Using provided data, generate a polite email response. End with a polite greeting.”
- Remember to always adjust the instructions based on your context
Before starting to work with Configurable AI and prompts we recommend to have a look on the prompt engineering guide by OpenAI for further instructions: https://platform.openai.com/docs/guides/prompt-engineering.

Prompt examples

Use the examples below as a starting point for configuring AI Actions. These examples provide prompts to configure according to your environment's needs - the attributes mentioned in some examples are shown as examples only, as the value of the attributes depends on which attributes are used and how.

Please note that no data is anonymized; the processed data includes only data selected within the action/prompt definition.

Remember that you can use the Actions to get an idea of what any data card in your ESM is about, like getting to the root cause of a ticket, understanding the status of a Change, or communicating the state of an identified Information security incident to non-technical stakeholders.

It is easy to adjust the prompt to your specific use case. Just add your instructions and remember to test often with real-life data.

Ticket Summary

Use Case: Provide a short summary of the selected ticket to allow quick decision making or simply provide it to recipients outside the service desk.

Prompt:

Ticket Classification

Use Case: Provide the classification for a ticket based on given definitions for Incident and Service request. Include the reason as well in the output.

Prompt:

Using key details from a service management support ticket, summarize the core issue, actions taken, causes identified, and current resolution status. Ensure the support agent understands the urgency, progress made, and next steps needed. Keep the overview clear and structured, without using introductory or concluding phrases, focusing solely on critical ticket information.
As such, you provide a crisp summary of 
{{Subject}}
{{DescriptionHTML}}
{{SolutionHTML}}

Data Definition: Activity (SPSActivityClassBase)

Attributes:

{{Subject}}
{{DescriptionHTML}}
{{SolutionHTML}}

Root Cause Analysis

Use Case: Help to identify the potential root cause of a ticket.

Prompt:

Using the provided service management data, analyze and identify the root cause
of the issue. Summarize key factors contributing to the problem and suggest the
most likely cause, supported by the data.
{{Subject}}
{{DescriptionHTML}}

Data Definition: Activity (SPSActivityClassBase)

Attributes:

{{Subject}}
{{DescriptionHTML}}

Create Solution

Use Case: Help to create a solution text for a closed ticket.

Prompt:

Using the service management data related to the ticket, generate a concise and
clear resolution text. Include the steps taken to resolve the issue, any relevant
troubleshooting actions, and the final solution applied. Ensure the text is suitable
for documentation and can be referenced for future similar issues.
{{Subject]}}
{{DescriptionHTML}}

Data Definition: Activity (SPSActivityClassBase)

Attributes:

{{Subject]}}
{{DescriptionHTML}}

KBA Draft

Use Case: Provide a draft for a Knowledge Base Article (KBA) based on the ticket description and solution.

Prompt:

As a Knowledge Manager, use provided service management data to create a
knowledge base article for Service Desk Agents. Include:
1. Title: Clear summary.
2. Overview: Issue intro from data.
3. Symptoms: Key indicators from data.
4. Troubleshooting: Steps and tools from data.
5. Resolution: Recommended fix.
6. Prevention: Best practices.
7. References: Related links.
Ensure clarity and actionability.
{{Subject}}
{{DescriptionHTML}}
{{SolutionHTML}}

Data Definition: Activity (SPSActivityClassBase)

Attributes:

{{Subject}}
{{DescriptionHTML}}
{{SolutionHTML}}

Asset summary of ticket assigned asset

Use Case: Aggregate asset related data of an asset linked to the specific ticket.

Prompt:

You are a virtual agent and provide a crisp summary of the data context provided. Each of the data points should start with a new line with the corresponding title.
Format the output as HTML in a bullet list. Take the titles from below (between // and :) and always style them bold.
Data Context:
// Name: {{T(SPSActivityClassIncident).asset.T(SPSComputerClassBase).Name}}
// Manufacturer: {{T(SPSActivityClassIncident).asset.manufacturerName}}
// Model: {{T(SPSActivityClassIncident).asset.Model}} 
// Serial Nr.: {{T(SPSActivityClassIncident).asset.SerialNumber}} 
// OS: {{T(SPSActivityClassIncident).asset.T(SPSComputerCLassOs).Manufacturer}} 
// OS Name: {{T(SPSActivityClassIncident).asset.T(SPSComputerCLassOs).Name}}
// Last Scan date: {{T(SPSActivityClassIncident).asset.T(SPSComputerClassBase).LastScanDate}}
// Order Date: {{T(SPSActivityClassIncident).asset.T(SPSComputerClassBase).T(SPSAssetClassOrder).orderDate}}

Data Definition: Activity (SPSActivityClassBase)

Attributes:

// Name: {{T(SPSActivityClassIncident).asset.T(SPSComputerClassBase).Name}}
// Manufacturer: {{T(SPSActivityClassIncident).asset.manufacturerName}}
// Model: {{T(SPSActivityClassIncident).asset.Model}} 
// Serial Nr.: {{T(SPSActivityClassIncident).asset.SerialNumber}} 
// OS: {{T(SPSActivityClassIncident).asset.T(SPSComputerCLassOs).Manufacturer}} 
// OS Name: {{T(SPSActivityClassIncident).asset.T(SPSComputerCLassOs).Name}}
// Last Scan date: {{T(SPSActivityClassIncident).asset.T(SPSComputerClassBase).LastScanDate}}
// Order Date: {{T(SPSActivityClassIncident).asset.T(SPSComputerClassBase).T(SPSAssetClassOrder).orderDate}}