AI Settings and Configuration

Overview

The Praetorian Guard Platform (PGP) provides granular control over AI capabilities through feature flags managed in the Organization Settings page. Each flag independently controls a layer of AI functionality, allowing administrators to progressively enable AI features based on their organization's comfort level and security requirements.

Feature Flags

AI capabilities are controlled by four feature flags, accessible under Settings → Organization for Praetorian administrators:

AI Inference

Property	Value
Flag	`disable_ai_inference`
Default	Enabled (flag absent)
Controls	All AI features including the AI Assistant, agent capabilities, and conversation endpoint

This is the master switch for AI. When AI Inference is disabled, the conversation API endpoint returns HTTP 403 and the AI Assistant UI is locked with a message: "AI inference has been disabled for this account." No AI operations of any kind can execute while this flag is active.

AI Agents

Property	Value
Flag	`enable_ai_agents`
Default	Disabled (must be explicitly enabled)
Controls	Customer access to Agent mode in the AI Assistant (Aurelius and all specialist agents)

This flag unlocks Agent mode for non-Praetorian users. Even with this flag enabled, users must hold an Analyst or Admin role to access agent functionality. Praetorian operators always have agent access regardless of this flag.

AI Automation

Property	Value
Flag	`enable_ai_automation`
Default	Enabled for new customers
Controls	Automated background agents (auto-triage, asset affiliation)

Controls whether passive background agents can automatically process data. These agents run without user interaction and handle tasks like finding triage and asset correlation.

AI Training

Property	Value
Flag	`disable_aiml_training`
Default	Disabled
Controls	Whether PGP can use account data for AI model training and improvement

When enabled, your organization's anonymized security data may be used to improve PGP's AI models. This is opt-in and does not affect the functionality of AI Inference or AI Agents.

Flag Interaction Matrix

AI Inference	AI Automation	AI Agents	Result
Off	Any	Any	All AI disabled. No AI Assistant, no agents, no automation.
On	Off	Off	AI Assistant in Query mode only. No agents, no automation.
On	On	Off	AI Assistant in Query mode + passive background agents (auto-triage, affiliation).
On	On	On	Full AI: Query mode, automation, and Aurelius agent mode with all specialist agents.

Role Requirements

Feature	Minimum Role	Additional Requirements
AI Assistant (Query Mode)	Any authenticated user	AI Inference enabled
AI Assistant (Agent Mode)	Analyst	AI Inference + AI Agents enabled
Auto-Triage Agent	System (automatic)	AI Automation enabled
Asset Affiliation Agent	Praetorian only	Manually triggered

Customer Type Restrictions

Freemium customers cannot enable AI Training or AI Inference. These flags are locked for freemium accounts. To access AI capabilities, organizations must be on a Managed, SaaS, Pilot, or Engagement plan.

Configuring AI Settings

Navigate to Settings → Organization
Scroll to the Praetorian Features section (visible to Praetorian administrators only)
Toggle the desired AI feature flags using the switches
Changes take effect immediately — no restart or deployment required

Each toggle provides instant UI feedback. If an error occurs during the flag update, the toggle automatically reverts to its previous state.