Security and Load Balancing Innovations in the Age of GenAI and Agentic AI

The AI/GenAI era is upon us. Gartner's Danielle Casey predicts that by 2028, 95% of organizations will have integrated GenAI into daily operations, up from 15% in 2025. This rapid growth in AI/GenAI, particularly in agentic AI workloads, presents considerable challenges for businesses. 

VMware vDefend lateral security and VMware Avi Load Balancer have been in lock-step with the AI trend, by (1) leveraging AI technologies to enable more value to customers and boost productivity and (2) addressing new requirements for AI workloads. There are three key aspects of AI as depicted in the diagram below: AI/ML, GenAI assistants and AI workloads. vDefend and Avi are ideally suited for AI for the following reasons:

• AI/ML: access to comprehensive data with rich context is critical for AI/ML to provide calibrated insights. Both vDefend and Avi have access to complete workload-level and platform-level data with context, due to their integrations with VMware Cloud Foundation (VCF) based modern private cloud.
• GenAI assistants: For a GenAI assistant to provide focused responses, comprehensive scope is critical in addition to having access to comprehensive data and context. vDefend and Avi’s multi-function / multi-layer software stack (versus disparate point products) enable GenAI assistants to provide high fidelity responses.
• AI workloads: AI workloads (including AI inference workloads and agentic AI workloads) are high performant, Kubernetes based and need to be load balanced and secured. vDefend and Avi are ideally suited because of their multi-terabit performance (due to software-defined scale-out architecture) and deep integration with Kubernetes workload, specifically vSphere Kubernetes Service (VKS) on VCF.

vDefend in the Era of AI/GenAI

AI/ML technologies in vDefend

At Broadcom, we've long leveraged the power of AI/ML within VMware vDefend to enhance cybersecurity defenses. Features like Security Intelligence for rule recommendations and Network Traffic Analytics (NTA) for detecting suspicious traffic are prime examples of how we leverage AI/ML to enhance lateral security solutions that make customer deployments easier, safer, and more cost-effective. 

GenAI Assistant for vDefend Firewall

Taking this further, we recently introduced Intelligent Assist for VMware vDefend, utilizing GenAI to turbocharge Threat Defense with GenAI-driven Intelligence. It simplifies and speeds up threat investigation and recommends mitigation options. We will extend GenAI assistant capability for the vDefend Firewall operations.

While popular use cases for GenAI assistants are using static data (documentation, configurations, technical support cases, etc), a unique aspect of vDefend’s GenAI assistants is the ability to provide insights to dynamic operations. These can be a real-time violation of security policy or an application being blocked due to a firewall rule. 

vDefend for AI Workloads

AI workloads, whether inference or agentic, have created a new attack surface. Securing AI workloads day-1 is paramount. 

At VMware Explore 2025, we are tech previewing zero-trust lateral security for AI workloads by extending VMware vDefend to VMware Private AI Foundation (PAIF) workloads running on VCF.  It highlights, for example, how interactions between AI workloads and shared services (network services, data sources, model repositories, etc.) can be locked down. With built-in support for VMware Kubernetes Service and 20 Tbps Distributed Firewall performance, lateral security can be realized for the most demanding AI environments..

VMware vDefend's declarative, tag-based policy model simplifies rule creation, accelerating not just the initial deployment but also the ongoing upkeep of security policies designed to handle rapid changes at scale demanded by the AI environment. Admins can create a segmented and secure AI workloads environment based on their functions (such as model runtime, agent builder, indexing and retrieval, etc.), and use vDefend to secure the communication between these critical components, ensuring integrity throughout the AI ecosystem.

Avi for the GenAI/AI Era

AI/ML Technologies in App Delivery and Web App Security

Avi delivers unparalleled application analytics with application health scores and end-to-end latency details for quick troubleshooting and performance insights derived from in-traffic logs. Its intelligence, powered by AI/ML, automates host selection, scales load balancers automatically, and optimizes application placement. Avi identifies and flags application anomalies, issuing alerts to prevent security threats. 

GenAI Assistant for Avi Dramatically Simplifies Operations

Customers are required to deliver applications at the speed of business while ensuring seamless performance and rapid issue resolution. Avi GenAI tool will enable organizations to consume and operate Avi in an intelligent and self-service manner. The solution integrates GenAI technologies with Avi advanced analytics. Through a conversational chat style interface, Avi administrators can instantly access AI generated explanations for Avi operations and application performance issues. to enable the following use cases:

• Knowledge Base (KB) Insights: Getting instant help from Avi public document with for quick answers to questions 
• Simplified Operations: Transitioning from manual steps to a data-driven methodology in Day 0 configurations and Day 1+ daily operations 
• Productivity Boost: Leveraging natural language reasoning and summarization to foster more efficient application performance monitoring and troubleshooting
• Recommendations: Taking the errors and stress out of tasks including upgrades with guided workflows and recommended best practices

Check out the tech preview demo to see these “GenAI for Avi” in action. 

Avi for AI Workloads

AI workloads require load balancing. Avi, with its integration with Kubernetes, multi-terabit performance (due to its software-defined architecture) and elastic scale-out operation is ideally suited to deliver load balancing to agentic AI workloads.

Avi is introducing a tech preview to extend load balancing to AI workloads on VMware Private AI Foundation (PAIF). Customers will be able to load balance AI workloads on Day 1. See a tech preview demo on how easy it is to deploy the LLM model endpoint and integrate with Retrieval-Augmented Generation (RAG).

Model Context Protocol (MCP) Intelligence in Avi

Avi will provide support for Model Context Protocol (MCP) with ability to maintain MCP persistence, integrate into MCP-based agentic AI ecosystems, and provide intelligence into MCP traffic with rich Avi analytics.

Here are three main use cases: 

• Avi load balancing MCP traffic with session persistence for reliable and scalable AI application deployments
• JSON Web Token (JWT) authorization support based on job roles so that app owners and operators can have different accesses to MCP tools
• Avi as a MCP Server to deploy load balancing as code for AI applications in four easy steps: 
  • Create a virtual service
  • Update the server pool
  • Update SSL certificates
  • Enable web app security

Watch these tech preview demos on Avi load balancing MCP traffic with session persistence and Avi as a MCP Server with Cursor.

VMware Explore Highlight Sessions

Join us at Las Vegas for VMware Explore 2025 and don’t miss the following AI-related sessions:

Editorial Notes: The information in this news release is for informational purposes only and may not be incorporated into any contract. There is no commitment or obligation to deliver any items presented herein.