Building an Agentic AI–Ready Data Platform on Nutanix with Dremio

Agentic AI introduces a new interaction model for enterprise data. Instead of humans carefully crafting queries and dashboards, AI agents now explore data autonomously—posing questions, refining queries, and acting on results in rapid iteration loops. This shift places new demands on the data platform: fast responses, elastic scale, clear semantics, and strong governance are no longer optional.

Nutanix addresses these demands by combining its infrastructure platform with Dremio as the analytics and semantic layer. Together, Nutanix Kubernetes Platform (NKP), Nutanix Objects Storage, and Dremio form a validated, full-stack data lakehouse designed for both modern analytics and the emerging requirements of agentic AI.

A Validated Nutanix Foundation for Dremio

At the core of this platform is NKP, which provides a scalable, Kubernetes® native  environment for running Dremio execution engines. NKP gives enterprises a consistent operational model for deploying, scaling, and managing Dremio across environments—on-premises, edge, or hybrid—using familiar Kubernetes constructs.

Beneath Dremio, Nutanix Objects Storage is validated as a high-performance primary storage layer, capable of sustaining the throughput and low-latency access patterns required by Dremio workloads, including Reflections. This validation matters because Dremio is not a lightweight BI tool; it is a distributed lakehouse query and execution engine that relies on predictable infrastructure behavior to deliver interactive analytics at scale.

Together, NKP and Nutanix Objects Storage provide the stable substrate on which Dremio can operate predictably—something agentic AI implicitly assumes.

Why Performance Matters More for AI-Driven Analytics

Dremio is often used to serve interactive, exploratory analytics, but agentic AI raises the bar further. AI agents typically operate in tight reasoning loops: generate a query, interpret the result, refine the question, and repeat. If any step is slow, the entire loop can degrade or fail.

This is where Nutanix Objects Storage becomes strategically important for Dremio. Objects Storage delivers both high throughput and low latency, enabling Dremio to scan large datasets efficiently while still responding quickly to repeated access of curated data. These characteristics are critical for sustaining AI-driven query workflows that demand responsiveness closer to human conversation speeds than batch processing timelines.

In short, Nutanix Objects Storage helps Dremio remain fast enough for agents to reason, not just query.

Open Lakehouse Analytics with Dremio and Iceberg

Dremio is built around open data standards, with Apache Iceberg serving as a foundational table format for lakehouse analytics. Iceberg enables snapshot isolation, schema evolution, and time-travel queries—capabilities that are essential for trusted analytics and AI governance.

Running Dremio on Nutanix Objects Storage allows Iceberg tables to scale without sacrificing performance. Metadata-optimized operations align well with object storage semantics, while sustained read throughput supports large analytical scans. For data engineers, this means fast ingestion and curation cycles; for AI systems, it means access to fresh, versioned data they can trust.

Dremio’s ability to optimize and query Iceberg tables efficiently is amplified by Nutanix Objects Storage’s’ performance characteristics, reinforcing the combined value of the platform.

Semantics: Where Dremio Enables AI Understanding

If Nutanix abstracts infrastructure, Dremio abstracts meaning. This becomes increasingly important as AI agents begin to interact with enterprise data directly.

Dremio’s AISemantic Layer allows teams to define trusted business views over raw data using virtual datasets, consistent metric definitions, and rich metadata. These semantics guide AI agents toward the correct datasets and calculations, reducing ambiguity at query time.

For agentic AI, this semantic layer is not a convenience—it is a prerequisite. Without it, agents may issue technically valid queries that produce logically incorrect or misleading results. Dremio provides the structured context that allows AI systems to reason about data consistently and responsibly.

Bringing Unstructured Data into Dremio-Driven Workflows

Enterprise data is no longer just rows and columns. Documents, reports, and images stored in Nutanix Objects Storage increasingly represent high-value inputs for AI systems.

With Dremio running on NKP and accessing Nutanix Objects Storage, analytics workflows can extend naturally into unstructured data. Capabilities such as AI_GENERATE allow Dremio to extract structured fields from text-based sources like PDFs and reports and expose them as relational datasets. Dremio can also apply AI-based classification and enrichment to unstructured image collections, creating metadata that can be queried and joined alongside traditional tables.

This allows Dremio to serve as a single analytical plane for structured and unstructured data—something agentic AI strongly benefits from.

MCP: Dremio as the Agent Interface to Data

As AI agents become more autonomous, they need a standardized, governed way to interact with enterprise data systems. Dremio’s Model Context Protocol (MCP) Server provides this interface.

Through MCP, Dremio exposes data discovery, schema inspection, and query execution in a form that AI agents can use programmatically. Within the Nutanix ecosystem, the Nutanix AI (NAI) connector for Dremio MCP connector enables NAI-managed model endpoints managed by the Nutanix AI solution to interface with Dremio’s MCP server, while still supporting other inference engines.

In this architecture, Dremio acts as the governed data access and execution layer for agents, while model endpoints focus on reasoning and inference. This separation is designed to keep enterprise controls—security, semantics, performance optimizations—enforced even as AI systems act autonomously.

Sustaining Agent Reasoning with Dremio Reflections on Nutanix Objects

Dremio’s Autonomous Reflections play a critical role in making agentic AI practical. Agents often generate complex, multi-table queries dynamically; waiting seconds or minutes for results disrupts reasoning chains and breaks conversational flows.

Dremio continuously observes query patterns and is designed to automatically materialize optimized Iceberg tables as Reflections. When stored on Nutanix Objects Storage, these Reflections benefit from fast, low-latency reads and scale-out throughput. Combined with Dremio’s Apache Arrow-based columnar execution engine, this enables AI agents to receive answers quickly enough to sustain iterative reasoning and decision-making.

Here, Nutanix Objects Storage and Dremio work together: storage performance enables query acceleration, and Dremio ensures that acceleration happens automatically.

Governance from Bronze to Gold with Dremio and Nutanix

While agents benefit from speed, enterprises require governance. Nutanix and Dremio support both governed ingestion and real-time data access without forcing a single approach.

Raw bronze data can be materialized into the lake on Nutanix Objects Storage and cataloged through Dremio’s Open Catalog, based on Apache Polaris. As data progresses through silver transformations, gold datasets can be exposed as curated virtual datasets and accelerated with Reflections. These gold layers become the trusted interfaces that BI tools, applications, and AI agents rely on.

With snapshotting, lineage, and time travel preserved across layers, enterprises retain full accountability—helping promote AI-driven outcomes to be auditable long after decisions are made.

A Nutanix Platform, Powered by Dremio

Together, Nutanix and Dremio deliver a data platform built for how AI actually works. NKP provides a resilient execution layer for Dremio at scale. Nutanix Objects Storage delivers high-performance, low-latency object storage optimized for analytical access. Dremio adds the semantic intelligence, governance, and acceleration that make data usable by both humans and AI agents.

The result is a platform that doesn’t just support high speed analytics, but enables agentic AI to operate safely, quickly, and at enterprise scale—on infrastructure you control.

©2026 Nutanix, Inc. All rights reserved. Nutanix, the Nutanix logo and all Nutanix product and service names mentioned are registered trademarks or trademarks of Nutanix, Inc. in the United States and other countries. Kubernetes is a registered trademark of The Linux Foundation in the United States and other countries. All other brand names mentioned are for identification purposes only and may be the trademarks of their respective holder(s).