RAI Platform Provisioning Options¶
Option 1: Managed Regional Endpoint¶
This provisioning option is exclusive to AOAI Service Teams because Responsible AI operates within the Azure OpenAI infrastructure. It provides a managed region deployment model, featuring a centralized endpoint. This approach simplifies the integration process and ensures high performance and compliance across the platform.
Benefits:
-
Centralized Management
- A managed endpoint consolidates all service interactions, reducing complexity in managing multiple endpoints.
- Centralized policy management allows the RAI team to handle updates, configurations, and monitoring seamlessly, ensuring smooth operations without additional overhead for AOAI Service Teams.
- Administrators gain better visibility and control over all integrated services, which enhances the ability to troubleshoot and optimize system performance.
-
Policy Consistency
- By sharing and enforcing consistent policies across all AOAI internal services, the platform minimizes discrepancies and ensures uniform compliance with Responsible AI guidelines.
- Streamlined operations reduce the risk of misalignment between teams and foster collaboration by adhering to the same operational and ethical standards.
- Unified policies simplify audits and regulatory reviews, as there is a single source of truth for policy implementation.
Option 2: Standard Azure Resource Creation¶
This option is designed to serve both AOAI Service Teams and AOAI Customer Teams, offering a more flexible approach by enabling users to create standardized Azure resources. These resources can be configured to meet specific requirements while remaining compatible with the RAI platform.
Benefits:
-
Customizability
- Users can define proprietary policies tailored to their unique operational or organizational needs. This ensures the platform can address a wide range of use cases, from general AI services to specialized solutions.
- Customized configurations provide the opportunity to integrate domain-specific rules or constraints, enhancing the overall utility and relevance of the deployment.
-
Flexibility
- Supports diverse deployment scenarios, from basic setups for individual services to complex configurations involving multiple teams and regions.
- Accommodates organizations with unique infrastructure or compliance requirements, allowing them to tailor resource allocation and governance policies.
- The ability to adjust configurations dynamically enables users to respond quickly to evolving business needs, such as scaling resources or implementing new AI models.
- By supporting a broad spectrum of use cases, this provisioning option empowers both internal teams and external customers to innovate while adhering to Responsible AI principles.
Comparison of Benefits¶
Managed Regional Endpoint | Standard Azure Resource Creation | |
---|---|---|
Option Detail | Azure RAI team provision and manage AACS resources. RAI policies also managed by RAI team. Customer need to wait RAI team response on resource provision and policy management. | Customer provision and manage AACS resource. RAI policies also managed by customers. Full control at customer hand. |
Ease of Management | High – Centralized management by RAI team | Moderate – Standard Azure resource provision. |
Policy Customization | Medium – Shared, centralized policies. Custoner can leverage DEV environment to test policy before request to add a production one. | High – Tailored policies per resource |
Flexibility | Medium – Only go to production need RAI team sign off. | High – Adaptable to diverse use cases |
Compliance Assurance | High – Uniform and consistent | High – Customizable to specific needs |