IN THIS ARTICLE

HallianAI is a powerful tool that requires careful administration to ensure effective use and security. This document outlines best practices for managing two distinct types of roles within the platform: Platform Capability Roles (what users can DO) and Data Access Roles (what data users can ACCESS). Together, these create a flexible, scalable permission model that supports both security and organizational agility.

Contents

<aside> <img src="notion://custom_emoji/f2cca387-5803-81f0-a7b9-00035b6fc5fc/20bca387-5803-802b-b24b-007aec271335" alt="notion://custom_emoji/f2cca387-5803-81f0-a7b9-00035b6fc5fc/20bca387-5803-802b-b24b-007aec271335" width="40px" />

Effective management of roles and permissions is crucial to ensuring the secure and efficient use of HallianAI. By separating platform capabilities from data access, organizations can ensure that their users have the necessary permissions to perform their jobs while protecting sensitive information and maintaining compliance with organizational policies and procedures.

</aside>

Roles and Permissions

HallianAI has several built-in roles with different permissions, including:

• Super Admin: Has full access to the platform, including settings, indexes, and user management.

• Application Administrator:

  1. Web scraper Administrator
  2. AI Assistant User
  3. Workflow Administrator
  4. Workflow User

• Settings Administrator: Can manage settings, including cloud configurations, AI assistant instructions, and index management.

  1. Dashboard Manager: Can view the Dashboard for HallianAI
  2. AI Assistant Instructions Manager: Allows management of Chat Settings for LLM Interactions
  3. Cloud Configurations Manager: Can configure and manage cloud resource settings
  4. Indexes Manager: Can create, update, and manage indexes, including uploading documents and updating AI assistant instructions.
  5. Model Deployments Manager: Can add, update, and manage model deployments.
  6. Roles Manager: Can create, update, and manage roles, including assigning users to roles.
  7. User Manager: Can add, update, and manage users.
  8. Public Chat Manager: Can enable or disable public chat, set the index and large language model used, and manage chat settings.
  9. Active Users Manager: Can view currently logged-in users and track user activity.
  10. License Manager: Can view license information and manage licenses.

  

Best Practices for Role Management

• Limit Super Users: The number of super users should be limited to a small number, typically system administrators who have access to the server, Azure, AWS, and Google Cloud.
• Departmental Roles: Create admin roles for each department, such as HR Admin, Finance Admin, and Engineering Admin, to give departments ownership and control over their areas.
• Power Users: Identify power users who can drive the use of models and model deployments for their departments. These users should have a higher knowledge of AI and how it works.
• Role-Based Access Control: Assign users to roles based on their job functions and responsibilities, ensuring that each user has the necessary permissions to perform their tasks.
• Granular Permissions: Use granular permissions to control access to specific features and functions within the platform.

Best Practices for Administrator Management

1. Admin Training: Provide admin training for users assigned to administrator roles, ensuring they understand the platform's features and functions.
2. Trust and Responsibility: Ensure that administrators understand the level of trust and responsibility that comes with their role, including the potential to delete or update sensitive information.