AI Governance

Introduction

Joget’s AI Governance framework provides a robust, multi-layered environment designed to balance enterprise innovation with strict operational control. It allows organizations to build and deploy AI-powered applications at speed, while ensuring full transparency, cost accountability, and safety throughout the entire AI lifecycle. The following are various monitoring tools at your disposal when working with AI in Joget:

AI Agent Builder

• Rule-Based Guard
• LLM Safety Check
• Webhook Guard
• Agent Execution Audit Trail

AI Designer

• AI Designer Governance

Monitor Menu

• AI Usage Dashboard
• Governance Health Check
• Application Performance Management

Core Governance Pillars

The framework ensures end-to-end security and management through three primary systems, each addressing a specific stage of AI development and execution:

• Design-Time Governance (AI Designer): Focuses on the development phase. It provides a governance console to track session activities, identify who is designing applications, and audit LLM token consumption, ensuring visibility before deployment.
• Runtime Governance (Agent Builder & AI Composer): Focuses on live execution. It employs content safety pipelines ("Guards") to filter input and output, alongside an extensive execution audit trail that logs every step of an agent’s behavior.
• Usage & Cost Governance (AI Usage Dashboard): Focuses on platform-wide analytics. This system aggregates all LLM API calls across the platform to provide granular visibility into token consumption trends, costs by model, and usage by specific plugins.

Summary of Governance Capabilities

Key Benefits

• Regulatory Compliance & Traceability: Maintains comprehensive audit logs, including user identification, node-level execution traces, and status reporting to meet compliance standards.
• Proactive Safety: Utilizes configurable Guards—including LLM-based, rule-based, and webhook-based filters—to block, sanitize, or redirect unsafe content in real-time.
• Cost Management: Prevents unexpected expenses by providing clear visibility into token usage patterns, allowing administrators to monitor and optimize LLM consumption across the platform.
• Operational Control: Enables administrators to monitor live processes and intervene immediately with the ability to abort running executions if necessary.