Citrix, a Cloud Software Group company, has announced the launch of NetScaler® AI Gateway, a specialized control layer designed to bring enterprise-grade governance and security to Large Language Model (LLM) deployments. As organizations shift AI from experimental pilots to mission-critical production, NetScaler AI Gateway provides the infrastructure to manage AI traffic with the same operational discipline, visibility, and cost control used for traditional web applications.
Core Launch: NetScaler AI Gateway for secure, high-performance LLM delivery.
Cost Management: Features token-based rate limiting and spillover routing to prevent budget overruns.
Performance: Implements token-latency-based load balancing across multiple LLM backends.
Data Protection: New integration with Protecto for agentic data classification and masking of sensitive info (PII) before it reaches the AI.
Agentic Support: The NetScaler Console MCP Server extends governance to agentic AI workflows via the Model Context Protocol.
Security: Includes WAF protections for LLMs and integration with Enkrypt AI for OWASP LLM threat detection.
Availability: NetScaler AI Gateway and NetScaler Console MCP Server are available now.
Traditional network infrastructure was not designed for the unique traffic patterns of AI, where costs are measured in tokens rather than just bandwidth. NetScaler AI Gateway acts as a unified control layer that eliminates the need for fragmented point tools. It allows IT teams to enforce consistent policies across all AI applications, ensuring that one "rogue" internal app doesn't exhaust the company's LLM token quota.
“AI is becoming embedded in the operational fabric of the enterprise,” said Steve Shah, General Manager of NetScaler at Citrix. “Before organizations scale AI across core business processes, they need trusted guardrails to manage risk, control cost, and maintain performance.”
A major highlight of the announcement is the integration with Protecto, an AI-native data protection platform. This partnership addresses the "speed vs. safety" dilemma by using agentic classification to detect and mask sensitive data in real-time. This ensures that while the AI receives the context it needs to be accurate, actual sensitive information is never exposed to the model, maintaining strict compliance and customer trust.
Amar Kanagaraj, Founder and CEO of Protecto, noted: "Enterprises can finally scale AI at full speed without putting their data at risk. Protecto’s agentic data classification catches what traditional tools miss."
NetScaler AI Gateway introduces several "AI-aware" routing capabilities:
Token-Based Rate Limiting: Enforces fair consumption across different departments.
Spillover Routing: If a primary model reaches its quota, traffic is automatically diverted to a secondary model to prevent service interruptions.
LLM Redaction: Intercepts and modifies prompts in transit to enforce governance and protect privacy.
AI Observability: Provides specific metrics on token usage, latency, and quota violations that can be exported to existing analytics platforms.
Beyond managing simple LLM queries, Citrix is preparing for the era of AI agents. The NetScaler Console MCP Server allows enterprises to securely connect AI agents to operational intelligence. By using a standardized Model Context Protocol (MCP) interface, Citrix ensures that autonomous agents operate within the same visibility and governance framework as the rest of the enterprise.
About Citrix
Citrix, a Cloud Software Group company, provides a secure app and desktop delivery platform that enables flexible work for companies of all sizes. By unifying legacy systems and AI agents, Citrix provides a single pane of glass to connect intelligence to execution across the business.
TechIntelPro empowers B2B technology decision-makers with the latest industry trends, deep insights, leadership perspectives and actionable intelligence to drive both organizational and professional milestones—and thrive on the tech pulse.