AI Deployment Service Models

Zoom's artificial intelligence architecture provides organizations with three distinct deployment models, each engineered to balance AI capability requirements with specific organizational security, compliance, and operational preferences. Understanding these deployment models enables administrators and decision-makers to select the optimal AI implementation that aligns with their business goals.

The following diagram provides an overview of Zoom's AI Deployment Service Models, detailed in the sections below. Diagrams across the AI Bluepaper typically default to the federated approach, but you can mentally substitute the relevant dataflows with your preferred deployment model. For instance, ZMO (Zoom-Hosted Models Only) customers can disregard the outbound Third-Party Model connections.

Zoom's Federated Approach: Maximum Flexibility Through Strategic Multi-Vendor AI Partnerships

The Federated Approach represents Zoom's most flexible AI deployment model, designed to deliver maximum functionality through strategic partnerships with leading AI service providers. This deployment model utilizes communication pathways between Zoom's platform and multiple best-in-class AI providers, creating an intelligent routing system that dynamically selects the optimal AI service for each specific task requirement.

Zoom’s AI Service Partners

Under the Federated Approach, Zoom communicates with multiple AI providers using APIs, including:

• Anthropic: Used for content generation and summaries.

• OpenAI: Used for content generation and summaries.

In addition to those services, Zoom also communicates with additional third-party providers for specific features, including:

• Perplexity AI Services: Deployed for real-time web-based research, current information retrieval, and dynamic fact-checking capabilities requiring internet connectivity.

• ElevenLabs: Used for voice generation.

This deployment model provides organizations with:

• Complete Feature Access: Full availability of all Zoom AI Companion-powered features without functional limitations.

• Intelligent Workload Distribution: Automatic routing of AI requests to the most suitable provider based on task characteristics and performance optimization.

• Maximum Flexibility: Ability to leverage specialized AI capabilities from multiple providers within a unified Zoom interface

• Scalability Optimization: Dynamic load balancing across multiple AI services designed to provide consistent performance during peak usage periods

Implementation Considerations for Enterprise Deployment

Organizations selecting the Federated Approach should evaluate data routing implications, as AI processing may occur across multiple third-party environments. This model is optimal for enterprises prioritizing maximum AI functionality and feature availability while maintaining standard enterprise security protocols.

Zoom-Hosted Models Plus (ZM+): Enhanced Privacy Through Amazon Bedrock Dedicated Instance Architecture

The Zoom-Hosted Models Plus (ZM+) represents a balanced deployment approach that combines extensive AI capabilities with additional data privacy control measures. Through Amazon Bedrock’s managed AI service infrastructure, Zoom provides access to dedicated instances of third-party language models. This approach enhances data control for organizations while preserving access to diverse AI capabilities.

This model utilizes Amazon Bedrock's managed AI service infrastructure to host Zoom-dedicated instances of third-party language models, and provide organizations with improved data control while maintaining access to diverse AI capabilities.

Technical Architecture and Service Delivery Model

Through Amazon Bedrock integration, Zoom maintains dedicated, isolated instances of select AI models, creating a hybrid approach that delivers:

• Zoom-Managed Model Instances: Data is processed on Zoom-managed instances where information is not sent to third-party model providers.

• Multi-Model Support: Access to multiple AI providers through Amazon Bedrock's managed service layer, though with potentially different model versions or configurations compared to direct federated access.

• Data Control: Processing occurs within Amazon's enterprise infrastructure, and is controlled by Zoom.

While ZM+ provides substantial AI capabilities, organizations should anticipate that some features available in the Federated Approach may have limited availability or modified functionality with ZM+.

Strategic Implementation Benefits for Privacy-Conscious Organizations

This deployment model serves enterprises requiring enhanced data privacy controls while maintaining access to advanced AI capabilities. Organizations benefit from improved data control without sacrificing the intelligent task routing and multi-model optimization that characterizes Zoom's AI approach.

Zoom-Hosted Models Only: Maximum Data Control Through Proprietary AI Infrastructure

The Zoom-Hosted Models Only deployment represents the most restrictive AI implementation option, exclusively utilizing a language model hosted privately within Zoom's dedicated infrastructure.

Technical Infrastructure and Service Limitations

This deployment model operates through:

• Zoom-Hosted Models Utilization: All AI processing conducted through Zoom's hosted models.

• Dedicated Infrastructure: Processing occurs entirely within Zoom's controlled environment without third-party AI service dependencies.

• Simplified Data Flow: Eliminates external AI provider data routing, creating the most straightforward data governance model.

Organizations selecting this deployment model should anticipate:

• Limited Feature Set: This model may result in a reduction of available AI features compared to Federated and ZM+ approaches.

• Reduced Capability Scope: AI functionality constrained to capabilities supported by Zoom's hosted model.

• Performance Variations: Response quality and task handling may differ significantly from multi-model approaches, particularly for specialized use cases.

Strategic Use Cases for Maximum Security Requirements

The Zoom-Hosted Models Only (ZMO) deployment serves organizations with stringent data control requirements, regulatory compliance obligations, or security policies that prohibit third-party AI service utilization. While providing the least flexible AI experience when compared to other deployment models, this approach offers simplified compliance documentation while maintaining the benefits of modern AI.

Generative AI Models Summary

The following tables summarize the differences between Zoom’s three generative AI service models.

Algorithms

Beyond generative AI capabilities, Zoom's platform also incorporates sophisticated task-specific artificial intelligence models (i.e., algorithms). These algorithms are designed to enhance core communication functions through automated processing and intelligent optimization, powering features like live transcription or personal audio suppression. These specialized AI systems operate independently of the generative AI deployment models, providing consistent functionality across all Zoom implementations regardless of selected LLM deployment approach.