Handling “Error 300001: Now LLM Guardian call not successful” in Now Assist GenAI Skill Logs

Prabu Velayutha · 3 weeks ago

Hi everyone,

I’m seeing repeated occurrences of the following error in sys_gen_ai_log_metadata for multiple Now Assist GenAI skills:

Now Assist Q&A
Resolution Notes Generation
Incident Summarization

The log entry shows:
“Error 300001: Now LLM Guardian call not successful.”

Has anyone encountered this issue before?
Looking for insights on what typically causes this error and how to troubleshoot or resolve it. Any guidance or experience would be greatly appreciated.

Thanks!

rpriyadarshy · 3 weeks ago

@Prabu Velayutha

Error 300001: “Now LLM Guardian call not successful” occurs when the Now Assist Guardian service, which is ServiceNow’s Responsible AI guardrail layer, fails while evaluating an LLM request for risks such as offensiveness, prompt injection, or sensitive content, and this failure blocks the entire LLM response even if the underlying model is healthy.

Because Guardian is a mandatory enforcement layer when enabled, any internal error, timeout, dependency failure, or platform-side issue in the Generative AI Controller or Guardian service causes ServiceNow to reject the request and surface this error.

Retrying the request or re-authenticating does not resolve the issue, since the failure occurs outside the LLM itself and before the response is accepted by the platform.

This error is commonly associated with configuration gaps, entitlement or role issues, unsupported or misrouted LLM providers, or temporary ServiceNow backend service disruptions.

Resolution typically requires validating Now Assist Guardian configuration and enablement, confirming AI Agent and Now Assist prerequisites, reviewing system logs for Guardian or GenAI Controller errors, or waiting for platform recovery if the issue is service-side.

If Issue still persist - Open a Support case.

Some More Insight on this FAQ-

Now Assist Guardian FAQ - ServiceNow Community

Regards

RP

bkeszl2 · 3 weeks ago

We started getting the same error after upgrading to Yokohama patch 11, updating the plugins associated with Now Assist in Virtual agent seems to have fixed the issue on sub-prod instances. Root cause is unknown.