The F5 AI Security Platform now incorporates an enhanced AI Gateway designed to support companies handling increased artificial intelligence inference workloads. The refreshed system integrates governance protocols, permission management and oversight capabilities for artificial intelligence frameworks, autonomous agents and applications within one platform.

The company presents the gateway as a control mechanism for addressing expenses, compliance, and protection challenges associated with expanded adoption of generative artificial intelligence across enterprise functions. Companies increasingly seek such oversight capabilities as they transition from experimental phases to operational implementation.

According to the company's 2026 State of Application Strategy Report, 77 per cent of organisations now view inference as their primary AI activity, with the typical organisation operating seven AI models. This transition holds significance as inference generates a continuous flow of queries, each carrying implications for costs, system performance, and data protection.

With the increasing adoption of artificial intelligence tools by organisations and the growing interconnection of software agents to data repositories and programming interfaces, enterprises encounter heightened complexity in establishing governance frameworks that determine model access permissions, usage parameters, and associated expenditures.

F5 reports that the gateway integrates three components: a Model Gateway for managing access and costs, an MCP Gateway for regulating agent-tool connections, and AI Guardrails for monitoring prompts and outputs. Financial parameters, network routing configurations, and permission frameworks may be administered from a central location and deployed throughout multiple operational settings.

The update emphasises enhanced monitoring of token usage, consistent with broader industry attention to AI workload cost structures. Large language models quantify text through units known as tokens, a measurement system directly linked to the cost structure of commercial artificial intelligence platforms.

The Model Gateway tracks token usage across providers, models, teams, and individual users, according to F5. The platform enables organisations to establish financial limits that apply in real-time during consumption rather than following the billing cycle.

The platform additionally directs queries among various models according to expense and guidelines, employing semantic caching alongside GPU-optimised load distribution to minimise costs. According to the company, the configuration aims to reduce token costs by as much as 60 per cent while maintaining compatibility with existing applications.

The solution also facilitates oversight of automated software agents, which organisations deploy with growing frequency to handle routine operations through connections to external applications and information repositories. As their adoption grows, organisations encounter questions regarding permission management and monitoring of these systems' activities on behalf of individual users or organisational units.

The company reported that its MCP Gateway establishes access restrictions, confining agents to designated resources such as APIs, data repositories, and retrieval-augmented generation frameworks. The system maintains a record of access details, including the time, the requesting agent, and the person or entity represented.

According to F5, the registry of approved MCP servers was designed to provide development teams with a consolidated reference point for accessible tools. This trend indicates a wider movement within the infrastructure supply sector to establish more structured governance over newly developing agent-driven software architectures.

The update also examines inputs and outputs during their exchange with artificial intelligence systems. Entities operating under regulatory frameworks report growing apprehension regarding potential unauthorised access to confidential data through third-party systems or attempts to circumvent protective measures.

According to F5, the AI Guardrails technology removes confidential information prior to model processing, prevents unauthorised prompt manipulation, and halts queries that pose evaluation risks. The platform additionally features monitoring records, capabilities for exporting security and event management information, and mechanisms governing where data is stored.

These mechanisms are designed to facilitate organisational oversight and regulatory adherence in industries where artificial intelligence systems process medical data, confidential details, and proprietary innovations. The company reported that the system meets SOC 2, ISO, and HIPAA compliance standards.

The more comprehensive F5 AI Security Platform had been unveiled previously as a system providing oversight, management, assessment, and operational safeguards for AI applications and their associated interfaces. F5 integrates the AI Gateway within its platform to connect oversight capabilities more directly with operational traffic flows.

F5 operates in a sector where vendors position themselves to connect businesses with an expanding array of both external and proprietary artificial intelligence systems. The organisation referenced findings from Gartner that characterise AI gateways as a growing component of AI architecture for major corporations managing regulated entry to models and MCP servers.

The chief product officer at F5 indicated that organisations are rapidly implementing artificial intelligence technologies while encountering difficulties in managing them effectively. Each artificial intelligence query has economic, security and governance ramifications, but companies predominantly depend on disconnected systems that only tackle a portion of these challenges. The outcome includes higher expenses, greater exposure to potential challenges, and more complex operations.

The F5 AI Security Platform incorporates the F5 AI Gateway, which offers a unified management interface for overseeing artificial intelligence deployments spanning multiple models, cloud environments, agents, and application systems. According to the executive, enterprises will require an intelligent control layer to manage artificial intelligence systems.

The company provides a platform that enables clients to advance technological development while retaining oversight, protection, and management capabilities.