The Challenge of Traditional Ticket Resolution
Traditionally, IT support agents have had to navigate through a multitude of knowledge articles and data sources to resolve issues. This process can be time-consuming and often frustrating, especially when dealing with complex or niche applications. Can Helix GPT Agent Assist help with this? The typical workflow involves:
- Receiving an initial inquiry from a user
- Searching through knowledge articles to identify potential solutions
- Communicating possible solutions back to the user
- Often, the initial inquiry is not specific enough, leading to a cycle of back-and-forth communication between the agent and the user to pinpoint the exact issue and resolution. This iterative process significantly extends resolution times.
Introducing Helix GPT Agent Assist
Helix GPT Agent Assist aims to address these challenges by providing a search interface that offers summarized responses to reported incidents. This technology is based on large language models (LLMs) like Azure Open AI, Open AI, and Google Vertex, which have been pre-trained with a vast amount of data. Helix GPT takes the power of these models and overlays your corporate data to provide relevant and contextual responses.
Here’s how it works:
- Data Source Management: Helix GPT pulls in data from various sources such as ticket data, log data, observability data, and knowledge sources like SharePoint and Confluence, as well as data lakes like Databricks or Snowflake.
- Contextual Responses: By overlaying raw data, the system responds based on prompts using the enterprise’s data and knowledge, not just generic information.
- Natural Language Understanding: The LLMs can understand natural language, logic, and reasoning to provide accurate responses.
Key Features of Agent Assist
- Summarized Responses: Instead of making agents sift through numerous knowledge articles, Agent Assist provides a summarized response from the most relevant articles.
- Conversational Approach: Agents can converse with the AI using natural language, allowing them to ask follow-up questions and refine their search.
- Referenced Sources: Responses are fully referenced to their original source documents, allowing agents to cross-reference information and ensure accuracy.
- Prompt Generation: The system includes out-of-the-box prompt generation capabilities, written in natural language, which can be customized.
- Enterprise Knowledge: The system is designed to securely use enterprise data, ensuring that no data is exported outside of the client’s infrastructure. The data is stored within the Helix platform’s vector store, and only information needed for a response is retrieved.
- Bring Your Own Model: Customers have the flexibility to use their own large language models, including open-source models or models from other AI vendors such as Azure or Google.
- Control: You can control how the system responds by using your corporate data and can fine-tune responses to target customer needs.
Do you want to see it in action?
Please see this video to see it in action – while the entire video is worth watching, the Agent Assist portion starts at 16:37 and goes till 30:06.
For more information, contact us.