Introduction

FortiAIGate serves as the central gateway between your AI applications and major large language model (LLM) providers such as OpenAI, Anthropic, and AWS Bedrock. Positioned at the core of your AI infrastructure, it enables organizations to deliver AI services efficiently while maintaining strict security controls over all LLM interactions.

FortiAIGate provides two primary capabilities:

• AI Flow handles intelligent AI application delivery, routing requests based on content and ensuring that traffic is processed securely and efficiently.

• AI Guard delivers comprehensive security aligned with the Open Worldwide Application Security Project (OWASP) LLM Top 10, offering protections such as prompt injection detection, data leak prevention, toxicity detection, and support for customizable security rules tailored to specific needs.

FortiAIGate includes a fully featured graphical interface that simplifies configuration and management. Administrators can easily set up AI Flow routing policies, define AI Guard security rules, and monitor real-time system activity. Detailed traffic logs and a visual dashboard provide full visibility into all AI-related requests passing through the system.

Built for modern environments, FortiAIGate runs as a containerized solution on Kubernetes, allowing the seamless deployment across public clouds, private clouds, or on-premise clusters. Its cloud-native architecture ensures scalability, portability, and operational consistency across diverse infrastructures.

This document provides the following information for FortiAIGate 8.0.0 build 0022.

Web browser support

Web browser

Google Chrome 111 and later Microsoft Edge 111 and later Mozilla Firefox 111 and later Apple Safari 16.4 and later Other browser versions have not been tested but might fully function. Other web browsers might function correctly but are not supported by Fortinet.

Prerequisites

The following are the minimum Kubernetes worker node resources allocated for all FortiAIGate containers.

Component	Requirements
vCPUs	4
RAM	16 GB
GPU	1× NVIDIA GPU with 24 GB VRAM
Local storage	1x 250 GB NVMe SSD
Kubernetes cluster RBAC requirements	The cluster-admin privileges are required.

Ensure that you have the following components before deployment:

• Kubernetes 1.25.0 or later—linux/amd64 or linux/arm64

• CNI Plugin—The Container Network Interface plugin must be installed and configured.

  • Common options: Calico, Flannel, Weave Net, and Cilium

  • Must support pod-to-pod communication

• kubectl—Ensure that your client can access the Kubernetes API server.

  • kubectl documentation

• Helm 3.10.0 or later

  • Helm installation guide

• Container registry ready and accessible

• Ingress controller deployed

• Optional: GPU nodes (if the GPU mode is required)

  • The supported GPU models include NVIDIA L4, NVIDIA A10, and NVIDIA A100.