The AI Registry

Stand: 25.05.2025

Overview

The AI Registry in Grace provides a structured, hierarchical way to register and organize key components of AI governance. This includes registering:

  • Use Cases (High-level applications)
  • Systems (Functional groupings of models)
  • Models (Technical implementations)

These components are interconnected and reflect the layered architecture of AI deployments in an organization.


Hierarchy of Registration

Grace uses a three-tier hierarchy to organize AI assets:

1. Models (Lowest Level)
  • Represent technical AI components, such as:
    • Classification models
    • Large Language Models (LLMs)
  • Models can be:
    • Assigned to one or multiple systems
  • Serve as the foundational building blocks for AI operations
2. Systems
  • Group together one or more models
  • Represent the technical or operational scope of a solution
  • A model may be reused in multiple systems
3. Use Cases (Highest Level)
  • Represent business-level applications of AI
  • Group one or more systems under a common purpose or function
  • A system may be linked to one or more use cases

Registration Options

In Grace, you can register use cases, systems, and models through:

1. Guided Flow (Wizard)
  • A user-friendly interface that walks you through the necessary inputs
  • Recommended for most users
2. API Integration
  • Suitable for bulk or automated registration of AI assets
  • Ideal for technical teams or organizations with large-scale deployments

Summary Table
Level Description Can Be Part of Multiple... Registered Via
Model Technical component (e.g., LLM, classifier) Systems Wizard or API
System Functional unit composed of models Use Cases Wizard or API
Use Case Business-level AI application Wizard or API