Skip to main content
The AI Assistant is a powerful tool that converts natural language descriptions into structured policy configurations.

Natural Language to Policy Conversion

Simply describe what you want your policy to do, and the AI will:
  1. Analyze your requirements
  2. Generate appropriate conditions
  3. Configure notifications
  4. Create a complete policy structure

Streaming AI Experience

When you submit a query, the AI processes it in real-time:
  1. Processing your request - Initial analysis begins
  2. Analyzing requirements - AI understands your intent
  3. Generating policy configuration - Policy structure is created
  4. Finalizing - Policy is ready for review
You’ll see progress indicators for each step as the AI works.

Two Creation Modes

Instructions Mode

Access via /assistant/instructions:
  • Enter natural language descriptions
  • AI generates complete policy configurations
  • Best for complex policies or when you want AI to handle the details

Conditions Mode

Access via /assistant/conditions:
  • Visual condition builder interface
  • More control over individual conditions
  • Best when you know exactly what conditions you need

Chat Interface for Policy Refinement

After the initial policy is generated:
  • Continue the conversation to refine the policy
  • Ask for modifications: “Make the limit $5,000 instead”
  • Request additions: “Also add a time window restriction”
  • The AI maintains context throughout the conversation

Policy Simulation and Testing

Test policies before deployment:
  • Use the “Simulate” tab in the sidebar
  • Enter sample payloads
  • See how the policy evaluates the payload
  • View evaluation traces and results

Policy Simulation

Learn about testing policies

Expression Editor (Payload Editing)

Advanced users can edit condition expressions directly:
  • Use the “Payload” tab to view and edit raw condition expressions
  • Modify expressions for fine-grained control
  • See how changes affect the policy structure

Thread Management and Conversation Continuity

  • Each conversation has a unique thread ID
  • Threads maintain context across multiple messages
  • You can return to previous conversations
  • Policy configurations are linked to their threads