Can Virtual Agent provide a ChatGPT-like conversational troubleshooting experience using AI Agents?
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
an hour ago
Hi Community,
We have a client requirement where they are looking for a more seamless, ChatGPT-like conversational experience within ServiceNow Virtual Agent.
We have demonstrated several Virtual Agent use cases using conversational topics and AI Agents. However, the client is looking for something more dynamic than predefined conversational flows.
For example, suppose an end user says:
“I’m not able to upload a document.”
The expectation is that the assistant should understand the issue and respond with relevant troubleshooting steps, for example:
1 Suggest a few troubleshooting steps.
2 Ask the user whether the issue was resolved.
3 If not resolved, continue the conversation based on the user’s response.
4 Ask contextual follow-up questions, such as the application being used, error message, file type/size, etc.
5 Based on the responses, investigate further using Knowledge, AI Search/RAG, skills, actions, APIs, etc.
6 If the issue still cannot be resolved, create/update an incident or transfer the conversation to a live agent with the relevant context.
In other words, the requirement is for the AI assistant to dynamically understand the conversation, maintain context, decide the next best question/action, and continue troubleshooting rather than simply triggering a predefined VA topic.
We are currently looking at Now Assist in Virtual Agent, AI Agents, Agentic Conversations and Agentic Workflows.
Any guidance or examples would be greatly appreciated.
Thanks!
Vishwamitra Manav
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
32m ago
Hi @VishwamitraM ,
Based on the requirement, I think Agentic Conversations in Virtual Agent would be a good fit.
You can still create custom VA topics and define a specific process when needed. The main difference is that with Agentic Conversations, you don't have to define every possible conversation path upfront. The agent can understand the user's request, maintain context, ask relevant follow-up questions, and decide which available topic, skill, action, Knowledge/AI Search, or AI Agent to use next.
So, if the requirement is to follow a fixed process → Custom VA Topic works well.
If the requirement is more like "I have an issue, figure out what information you need, troubleshoot it, and decide what to do next" → Agentic Conversations in Virtual Agent, skills, and actions would be a better fit.
Go through this document to understand more about Agentic conversations in Virtual Agent:
-------------------------------------------------------------------------------------------------------------------------------------------
If my response solves your query, please mark it as helpful by selecting accept as Solution and Helpful. Let me know if anything else is required.
Thanks,
Prerna
