Hi everyone,
I’m trying to expose a custom AI Agent within ServiceNow’s Virtual Agent on the portal, but I’m running into an issue related to LLM provider configuration.
👉 Setup:
In AI Agent Studio, I’ve configured Azure OpenAI as the default LLM provider.
In Virtual Agent Designer, topics use Now LLM as the default provider.
In Now Assist Admin, the LLM provider setting aligns with Virtual Agent (NowLLM).
🚩 Problem:
With Now LLM selected in Now Assist Admin:
Virtual Agent topics are detected and work.
- In my AI Agent, “Display in Virtual Agent” under availability tab is not visible.
My custom AI Agent is not detected through Virtual Agent, when triggered—it returns:
“Unfortunately, there is not enough information available to address this query completely…”
This is for the VA Topic below:This is for the AI Agent Below
With Azure OpenAI selected:
My AI Agent works within Virtual Agent.
But built-in Virtual Agent topics fail to trigger or be detected. Everytime the AI Agents gets triggered even though other topics exist.
VA Topic prompt is starting the AI Agent:When Azure OpenAI is selected:
I need both topic types to coexist:
Custom AI Agent using Azure OpenAI.
Default VA topics powered by Now LLM.
These both should run together in the same Virtual Agent simultaneously.
❓ Questions:
Is there a way to configure the Virtual Agent or Now Assist settings to support dual LLM providers—Azure OpenAI for my AI Agent and Now LLM for standard topics?
Alternatively, is it possible to specify or override the LLM provider for individual topics or agents, similar to how AI Agent Studio allows provider selection?
Or is the best practice to use workaround approaches—like subflows, agentic workflows, or Now Assist Skill Kit—to bridge between the two LLM ecosystems?
Any guidance, documentation references, best practices, or configuration tips would be greatly appreciated.
Thanks in advance!
