Kathy Gibson reports from Nutanix .Next in Washington DC – Technology has always been a fast-moving industry, but the rate at which artificial intelligence (AI) is developing and being adopted puts most other innovations in the shade.
While AI itself was many years in the making, generative AI (GenAI) put the technology on the map, and dramatic advances are now seeing the rapid emergence of agentic AI as well.
“And agentic AI is finding use cases very quickly,” says Rajiv Ramaswami, president and CEO of Nutanix. “We are helping to make it a reality, with customers deploying it in a safe and simple manner.”
While AI can help organisations improve their processes, drive up productivity, and increase their competitiveness, its implementation can be fraught with challenges. And a poorly-executed solution could dangerously expose intellectual property (IP) and data.
Lee Caswell, senior vice-president: product and solutions marketing at Nutanix
“For many customers, they simply don’t know where to begin on their AI journey,” says Lee Caswell, senior vice-president: product and solutions marketing at Nutanix.
Apart from the up-front concerns around IP and data, geographic repatriation is becoming a major issue as customers look to classify data and repatriate it to sovereign clouds.
As they grapple with the many concerns, customers are looking for more than building blocks, Caswell says. “They want to be able to deploy turnkey solutions.”
Last year, Nutanix introduced the concept of GPT in a Box as its initial focus for Nutanix Enterprise AI, focusing on downloading models.
“As the year progressed, we saw AI moving at a dramatic clip,” Ramaswami says. “As such, we have come a long way in working with Nvidia and today we are announcing our full support for the Nvidia AI stack – NIM, Llama, Nemotron Reasoning, NeMo Retriever and NeMo Guardrails.
“This will allow Nutanix to provide agentic AI that offers simplicity, security, and control that is efficient.”
Benefits for customers include the fact that they can deploy agentic AI using their own data – with their choice of model and environment. They will be able to share applications. And they will be able to rely on a predictable cost when running in the cloud.
“We have been working very closely with Nvidia on their Nvidia Inferencing Microservices (NIM), and have already been certified with Nvidia for their Enterprise AI and unified storage offerings,” Caswell adds.
“We see Nvidia as a path to bring AI to the enterprise that takes out all the enterprise complexity.”
Nutanix today announced the general availability of the latest version of the Nutanix Enterprise AI (NAI) solution, adding deeper integration with Nvidia AI Enterprise including Nvidia NIM microservices and the Nvidia NeMo framework, to speed up the deployment of Agentic AI applications in the enterprise.
NAI is designed to accelerate the adoption of generative AI in the enterprise by simplifying how customers build, run, and securely manage models and inferencing services at the edge, in the data centre, and in public clouds on any Cloud Native Computing Foundation (CNCF)-certified Kubernetes environment.
The latest NAI release extends a shared model service methodology that simplifies agentic workflows, helping to make deployment and day two operations simpler. It streamlines the resources and models required to deploy multiple applications across lines of business with a secure, common set of embedding, reranking, and guardrail functional models for agents. This builds on the NAI core, which includes a centralised LLM model repository that creates secure endpoints that make connecting generative AI applications and agents simple and private.
“Nutanix is helping customers keep up with the fast pace of innovation in the Gen AI market,” says Thomas Cornely, senior vice-president of product management at Nutanix. “We’ve expanded Nutanix Enterprise AI to integrate new Nvidia NIM and NeMo microservices so that enterprise customers can securely and efficiently build, run, and manage AI Agents anywhere.”
“Enterprises require sophisticated tools to simplify agentic AI development and deployment across their operations,” says Justin Boitano, vice-president of enterprise AI software products at Nvidia. “Integrating Nvidia AI Enterprise software including Nvidia NIM microservices and Nvidia NeMo into Nutanix Enterprise AI provides a streamlined foundation for building and running powerful and secure AI agents.”
NAI for agentic applications can help customers:
- Deploy Agentic AI Applications with Shared LLM Endpoints – Customers can reuse existing deployed model endpoints as shared services for multiple applications. This re-use of model endpoints helps reduce usage of critical infrastructure components including GPUs, CPUs, memory, file and object storage, and Kubernetes clusters.
- Leverage a Wide Array of LLM Endpoints – NAI enables a range of agentic model services including Nvidia Llama Nemotron open reasoning models, Nvidia NeMo Retriever and NeMo Guardrails. NAI users can leverage Nvidia AI Blueprints which are pre-defined, customisable workflows, to jumpstart the development of their own AI applications that leverage Nvidia models and AI microservices. In addition, NAI enables function calling for the configuration and consumption of external data sources to help AI agentic applications deliver more accurate and detailed results.
- Support Generative AI Safety – This new NAI release will help customers implement agentic applications in ways consistent with their organisation’s policies using guardrail models. These models can filter initial user queries and LLM responses to prevent biased or harmful outputs and can also maintain topic control and jailbreak attempt detection. For example, Nvidia NeMo Guardrails are LLMs that provide content filtering to filter out unwanted content and other sensitive topics. These can also be applied to code generation, providing improved reliability and consistency across models.
- Unlock Insights From Data with Nvidia AI Data Platform – The Nutanix Cloud Platform solution builds on the Nvidia AI Data Platform reference design and integrates the Nutanix Unified Storage and the Nutanix Database Service solutions for unstructured and structured data for AI. The Nutanix Cloud Infrastructure platform provides a private foundation for Nvidia’s accelerated computing, networking, and AI software to turn data into actionable intelligence. As an Nvidia-Certified Enterprise Storage solution, Nutanix Unified Storage meets rigorous performance and scalability standards, providing software-defined enterprise storage for enterprise AI workloads through capabilities such as Nvidia GPUDirect Storage.
NAI is designed to use additional Nutanix platform services while allowing flexible deployments on HCI, bare metal, and cloud IaaS. NAI customers can also leverage the Nutanix Kubernetes Platform solution for multicloud fleet management of containerised cloud native applications, and Nutanix Unified Storage (NUS) and Nutanix Database Service (NDB) as discrete data services, offering a complete platform for agentic AI applications.
NAI with agentic model support is generally available now.
An AI case study
A major lifestyle retailer with a footprint across the US, Tractor Supply Co (TSC), is an early AI pioneer and today uses AI to empower staff members and provide a frictionless experience for customers.
Glenn Allison, vice-presdient Tractor Supply Co (TSC), explains that every team member in the group is AI-enabled. “We have deployed GenAI to team members, giving them access to product information, recommendations, and other information that they can use to help customers,
“And we have recently expanded the access to information by adding multiple large language models (LLMs).”
Most recently, partnering with Nutanix, TSC has deployed edge computing in hundreds of its stores.
“This has allowed it to add storage and computer vision AI models to detect customers’ experiences and identify opportunities to enhance our service. It opens up sales opportunities as well.”
TSC runs a cloud-smart architecture on multiple data centres, with the ability to burst into the cloud when necessary.
“We are also enabling cloud-based analytics platforms, that will allow us to accelerate and deliver new capabilities,” Allison says.