Learn how Now Assist helps defend against AI-specific threats including offensive content, prompt injection, and sensitive subject detection using Now Assist Guardian.

Now Assist Guardian monitors requests sent to large language models and their responses to protect you, your users, and your data. It detects offensive or harmful content, prompt injection attempts, and sensitive subjects, and can log or block detected content depending on your configuration.

Configure guardrails

The following topics describe how to set up Now Assist Guardian and configure the guardrails available for Now Assist skills and AI agents.

Learn how Now Assist Guardian monitors generative AI content at runtime, what categories of content it detects, and how logging and blocking work.

Manage the Guardrail service provider used by Now Assist Guardian.

Turn on offensiveness protection to log and optionally block offensive content in AI-generated text and conversations.

Activate or deactivate prompt injection attack protection for AI-generated text and conversations.

Set up filters to redirect users to a different topic when certain subject material is detected in a Virtual Agent conversation.

Enable Now Assist Guardian for AI agents

Enable Now Assist Guardian in AI agents to automatically identify and block offensive messages, helping protect your agentic workflows from harmful content.

Monitor guardrail activity

The following topics describe how to review and export Now Assist Guardian logs to evaluate guardrail effectiveness and support security review.

Now Assist Guardian analytics

Monitor the performance of guardrails enabled through Now Assist Guardian, including tracking how often offensive content and prompt injection attempts are detected.

Export logs from Now Assist Guardian to get insights into how often different guardrails are being detected and used.