Self Service Knowledge Base Assistance

The IT Support Agent uses advanced natural language processing, semantic search, and contextual understanding to interpret user queries and retrieve the most relevant knowledge content. The agent guides users toward self-resolution by intelligently recommending articles, highlighting key solutions, and, where required, escalating unresolved issues toward incident creation or human support.

The IT Support Agent for Knowledge Article Retrieval performs the following:

Natural Language Query Processing

The agent accepts user questions and issues typed in conversational, everyday language. Using AI-driven interpretation, the support agent understands user intent even when technical terminology or precise keywords are not used.

Example:
User: “My WiFi keeps disconnecting suddenly. How can I fix this?”
Agent: Interprets the problem as a network connectivity issue and begins knowledge retrieval by providing the solutions.

To execute the above, perform the following steps:

Log in to the Application as an End User.
Click on the AIW chat, and enter the required prompt as a natural query. The support agent processes the query and provides solution as required.

Figure: User Query in Conversational Language

Article Retrieval and Presentation

For each user issue, the support agent retrieves the top 3–5 relevant knowledge articles from the knowledge management system.

To execute the scenario, perform the following steps:

Log in to the Application as an End User.
Enter your query or issue description using plain language in the AIW chat.

Figure: Search for KB articles
The IT Support Agent analyzes the query and determines the most relevant topics.
The agent presents a selection of 3–5 knowledge base articles, each as a clickable link including a concise, user-friendly summary or key steps.

Figure: Agent response

Key Solution Highlighting and Escalation

If the user indicates that the recommended article(s) did not resolve the issue:

The agent provides additional solutions, including step-by-step instructions or a curated troubleshooting guide, with sources from the web.
The agent asks the user to confirm if the new recommendations were successful.
After multiple unsuccessful solution attempts, the agent proactively offers to escalate the issue—by either raising an incident or connecting to a Live Agent (LAT).

To execute Step-by-Step Solutions and Escalation Prompt, perform the following steps:

Log in to the Application as an End User.
Enter prompt, as “unable to connect to VPN”.
Agent responds with a set of Knowledge Articles and enquires if the articles were helpful to resolve the issue.

Figure: User Prompt
Based on the user response, agent further enquires till the issue is addressed.
Figure: AIW Response
Based on the response, to the agent provides further solution if the user is not satisfied.

Figure: AIW Response
After the user confirms that the solution is not working, it further provides more solutions to resolve the issue and finally asks if the issue can be escalated further.

Figure: Agent response to Escalate
Upon confirmation, the Agent enquires if an Incident can be created to address the issue.

Figure: Agent response to create Incident
On confirmation, agent creates an incident and provides ID further tracking.

Figure: Incident

Sample Prompts

Sample prompts for the Key Solution Highlighting and Escalation flow, as the agent interacts with the user after initial knowledge articles have not resolved the issue are as follows:

If User Indicates the Articles Did Not Help:

Agent: “Sorry those articles didn’t resolve your issue. Let’s try a different approach. Here’s a step-by-step guide that might help, sourced from our support community:
[Step-by-step instructions]
Did any of these steps fix the problem?”
Agent: “Thanks for letting me know those solutions weren’t effective. Here’s another troubleshooting guide, with detailed steps and screenshots:
[Guide link or summary]
Were you able to resolve the issue using this guide?”

Agent Asks for Confirmation:

Agent: “Did the solution I provided above help resolve your issue?
- Yes, it worked
- No, I still need help”
Agent: “Can you confirm if you were able to fix the problem after following those steps?”

If Multiple Attempts Are Unsuccessful—Agent Proactively Offers Escalation:

Agent: “I see the issue is still not resolved after several attempts. Would you like me to connect you with a live support agent for further assistance?”
Agent: “Since the self-service solutions haven’t worked, I can raise an incident ticket for you. Would you like to proceed with opening a support request?”
Agent: “Would you prefer to continue troubleshooting, speak to a live agent (LAT), or have me create a support ticket so this can be reviewed by IT?”

Graceful Handling of No Matches

When the agent does not find relevant articles:

The user is notified with a friendly message that no suitable solutions were found.
The agent suggests next steps, such as rephrasing the query or requesting further support.
The user may be offered the option to create an incident or escalate to a Live Agent.

User: Provide Knowledge Article For VM Provisioning
Agent: Gracefully handles the articles that are not present in the Knowledge Base and enquires if further help is required.

Figure: Graceful handling unavailable articles

Seamless Redirection to Incident Creation

If the self-service process does not resolve the user’s issue or at the user’s explicit request:

The agent offers to create a new incident, carrying all relevant context (the user’s original issue and attempted solutions).
The user is asked to confirm the creation of an Incident.
Upon confirmation, the agent submits an incident and provides a hyperlink to the incident record for easy tracking.

To execute Incident creation redirection, perform the following steps:

Log in to the Application as an End User.
Enter prompt to get the issue resolved.
Example: “Snagit tool is not working”.

Figure: User Prompt
Support Agent further requests for more information from the user.
Based on the user input, it further suggests more steps to resolve the issue.

Figure: Resolution Steps by Agent
The Agent will converse with the user to try more steps to resolve.
If the user is not able to resolve then the agent creates an incident.

Figure: Incident Creation