Dremio’s latest release delivers AI-Driven Intelligence with the Agentic Lakehouse

Companies are racing to operationalize agentic AI, yet the final process of getting from data to decision is extremely difficult,  requiring data integration, tuning, and governance management. With Dremio’s latest release, we remove these blockers by putting natural‑language intelligence, explainability, and self‑optimizing performance directly into the lakehouse experience. You get clarity and control without copies, lock‑in, or manual knobs.

In this latest release, Dremio Agentic Lakehouse deliver’s an integrated AI agent, AI-native SQL functions,  a first-class VS Code experience and Service Users to your lakehouse environment. This enables companies to move from questions to actions faster, with governance and control.

Dremio’s AI Agent

Dremio’s AI Agent changes how people and systems interact with data by turning natural-language questions into insight, and  answers. Instead of writing SQL or switching between tools, users simply ask a question and immediately receive a clear visualization or an explanation they can further explore and refine. The agent interprets intent, maintains context across follow-ups, and adapts as the conversation evolves—guiding you from understanding what happened to determining what should happen next.

Behind the scenes, Dremio’s AI Agent leverages Dremio’s AI Semantic Layer and governance framework to ensure every response reflects the right definitions, the right access controls, and the right level of detail. This creates a trustworthy, repeatable interaction model where insights are consistent across teams and governed by your organization’s data policies.

By compressing the full Questions → Insights → Actions cycle into minutes, Dremio’s AI Agent enables deeper exploration, accelerates decision-making, and unlocks more value from your Dremio Agentic Lakehouse.

AI Functions

Dremio’s AI Functions bring LLM power straight into SQL so you can work with unstructured data in your Dremio lakehouse without new tools or pipelines. Use AI_GENERATE to pull structured fields from documents or images (e.g., invoice number, due date), AI_COMPLETE to summarize or draft short explanations you can share, AI_CLASSIFY to label text by sentiment, topic, or type, —turning unstructured data into clean tables and columns. By eliminating the need for extra pipelines or tools, AI Functions allow unstructured data to live directly in the lakehouse with structured data. This streamlines analysis and speeds up access to insights.

VS Code Extension

With Dremio’s new VS code extension capability you can work end‑to‑end in VS Code. It is now easy to  browse the Dremio Open Catalog, write and run SQL with autocomplete and formatting, and view results—all without context‑switching. Built‑in Microsoft Copilot support lets you discover datasets and generate SQL using natural language, accelerating both onboarding and expert workflows.

Service Users 

Dremio introduces Service Users, purpose-built identities for machine-to-machine access across applications, pipelines, and AI systems. These non-human accounts authenticate using OAuth client credentials and operate through API-first interfaces like REST and Arrow Flight—eliminating the risks of shared credentials or long-lived tokens.

Because Service Users can’t access the UI and can be scoped to precise roles and policies, they significantly reduce attack surface while integrating cleanly with Dremio’s full governance model, including row-level filtering and column masking. Whether powering CI/CD pipelines, data integrations, or cross-platform Iceberg workflows, Service Users provide a secure, modern foundation for automated interaction with your Dremio lakehouse environment.

Delivering the Agentic Lakehouse

Together, these enhancements in this release strengthen Dremio’s vision for the Agentic Lakehouse—one built on an AI-enabled semantic layer that enhances discoverability and trust, autonomous optimization that continually elevates performance without manual tuning, and an open, hybrid Iceberg architecture that delivers flexibility across on-premises and cloud environments without locking you in.