Nutanix launches Agent Gateway to govern AI agents
Fri, 3rd Jul 2026 (Today)
Nutanix has made Nutanix Agent Gateway generally available as part of Nutanix Enterprise AI 2.7.
The software is designed to manage interactions between AI agents, large language models and enterprise tools through a single control point. It is aimed at AI developers and platform teams that need to govern agent activity, manage access policies and monitor token consumption across agentic AI deployments.
The launch comes as companies move from AI trials to wider deployment of autonomous software agents that interact with business applications, internal data and external models. That shift has increased scrutiny of governance, access security and model usage costs, particularly where organisations use multiple providers or a mix of public cloud and self-hosted systems.
Nutanix Agent Gateway is integrated with the company's Enterprise AI platform and can be used across environments that rely on frontier models hosted in the public cloud, as well as private models run by customers.
Control layer
Nutanix describes the software as a control layer between requestors, including AI users and agents, and AI models and Model Context Protocol servers. It applies access control policies and tool-level filtering across agents, aiming to let them use enterprise resources within a governed environment.
The software includes governance controls for Model Context Protocol servers, unified observability for token usage and model activity, audit logs for requests and a unified application programming interface for access to both external and self-hosted models. It also includes token-based rate limiting to enforce quotas and give teams visibility into usage by agent and team.
A central element of the launch is cost oversight. The software centralises token observability across model providers, giving IT and platform teams a way to monitor usage, allocate costs and track AI spending more closely.
That visibility could also help companies decide which workloads to move to self-hosted models rather than continue relying on external services. For organisations trying to balance performance, governance and spending, that decision has become more important as AI use expands beyond small pilot projects.
Regional demand
Nutanix linked the product to rising demand in Asia Pacific and Japan, where businesses are expanding AI use but facing pressure to put stronger controls around deployments. The challenge is becoming less about proving AI tools can work and more about managing how they are used in day-to-day operations.
One issue for technology teams is that agentic AI systems do not call only a single model. In practice, agents may connect to several models, internal tools and private data sources as they complete tasks, making oversight more difficult and spreading costs across different providers.
In that context, centralised policy management and audit trails are becoming more important for companies in regulated sectors or those handling sensitive internal information. Nutanix said its gateway records every Model Context Protocol request as part of a broader audit trail for AI governance.
Commenting on the launch, Daryush Ashjari, Chief Technology Officer & Vice President, Solution Engineering, Asia Pacific and Japan, Nutanix, said: "Across APJ, we're seeing AI adoption accelerate at a pace that governance simply isn't keeping up with. As organisations move beyond experimentation and start scaling agentic AI, the problem quickly shifts from model performance and accuracy to visibility and control, specifically, understanding and governing how these agents are used and behave in real-world environments. Nutanix Agent Gateway addresses this head-on by introducing a consistent control plane for agentic AI. It gives IT leaders the visibility and governance they need to secure, manage, and scale AI agents with confidence, irrespective of where the underlying models are running."