AI Functions Power Faster Agentic Analytics and Insights

The rapid growth of the use of AI throughout the modern data stack has transformed how organizations extract insights from their data. With our latest release, we're excited to announce the general availability of AI Functions — a capability that brings the power of Large Language Models (LLMs) directly into SQL execution, making Dremio’s Agentic Lakehouse the leading solution for AI-enhanced agentic analytics.

Key highlights:

• Agentic analytics is the next evolution of business intelligence, where AI agents autonomously execute analytics workflows from data preparation to insight generation.
• True agentic performance begins at the lakehouse, which unifies, governs, and accelerates how data is accessed and analyzed.
• AI Functions enable business users to embed intelligence directly in SQL, eliminating pipelines and unlocking real-time analysis of unstructured data.
• Dremio delivers the enterprise foundation for agentic AI insights, combining open lakehouse architecture with native AI execution for faster, governed insights.

What is agentic analytics?

Agentic analytics represents the next evolution of business intelligence, where AI agents autonomously execute analytics workflows—from data discovery and preparation to insight generation and visualization. Unlike traditional BI tools that require manual queries and human interpretation, agentic analytics enables AI agents and business professionals to ask questions in natural language and receive fast, accurate answers backed by unified, governed data.

Dremio's approach to agentic analytics is fundamentally different from competitors. While other platforms require data copies, complex pipelines, or lack proper governance, Dremio's Agentic Lakehouse unifies data where it lives, enforces governance at every layer, and provides business context through the AI Semantic Layer. This means AI agents can deliver trustworthy insights instantly—without ETL, vendor lock-in, or operational overhead.

Why enterprise agentic analytics starts with the lakehouse

True enterprise agentic analytics requires more than just AI capabilities—it demands a unified, governed data foundation that enables AI agents to operate securely, accurately, and at scale. Dremio's Agentic Lakehouse provides this foundation, combining open lakehouse architecture with native AI execution to accelerate AI implementation while reducing cost and complexity.

• Unifying data access for human and AI agents: Dremio federates queries across all your data sources with zero ETL, enabling both business teams and AI agents to access consistent, up-to-date information without data duplication or pipelines.
• Ensuring semantic consistency at the data layer: The AI Semantic Layer gives AI the business context required to find the right data and deliver accurate insights, eliminating ambiguity and ensuring consistent interpretations across all analytics workflows.
• Embedding governance into every analytic action: Fine-grained access controls and lineage tracking ensure that AI agents operate within security boundaries, maintaining compliance while processing sensitive data in your secure lakehouse environment.
• Enabling extensibility through open architecture: Built on Apache Iceberg, Polaris, and Arrow, Dremio enables agent choice—use integrated agents or bring your own with MCP—without vendor lock-in or proprietary formats.
• Delivering performance through data proximity: Autonomous Reflections and Automatic Iceberg Clustering optimize queries automatically, delivering sub-second performance at a fraction of the cost of traditional data warehouses or cloud platforms.

Transforming unstructured data into powerful agentic AI insights

Organizations today face an unprecedented challenge: extracting meaningful insights from large volumes of unstructured data that enterprises store. Previously, analyzing customer feedback, processing documents, or synthesizing text data required complex multi-step pipelines, specialized tools, and often manual intervention. AI Functions eliminate these barriers by enabling direct LLM prompting during SQL execution, accelerating time to insight. 

Consider a retail organization analyzing thousands of customer reviews stored in its lakehouse. What once required a dedicated data pipeline to classify sentiment, extract key themes, and summarize feedback can now be accomplished with a single SQL query. This transformation reduces operational complexity while enabling agentic AI insights from unstructured content.

AI Functions: agentic intelligence without pipelines

Dremio AI Functions represent a fundamental shift in how organizations analyze unstructured data. By embedding AI directly into SQL execution, AI Functions eliminate the need for separate pipelines, external ML tools, or data duplication—delivering the fastest path to AI insights while reducing cost and operational overhead.

Embed AI directly into SQL workflows

Traditional approaches to AI-powered analytics require complex architectures: data extraction pipelines, separate ML infrastructure, model deployment and monitoring, and custom integration code. AI Functions eliminate this complexity by bringing LLM capabilities directly into SQL execution. Business analysts and AI agents can now query unstructured data using familiar SQL syntax, without waiting for data engineering resources or building specialized pipelines.

This native integration means insights are delivered faster, data stays governed within your lakehouse, and operational overhead disappears. Whether you're processing customer feedback, analyzing financial documents, or extracting insights from call logs, AI Functions enable analysis where your data already lives—with zero ETL and no vendor lock-in.

Enterprise value:

• Accelerate time-to-insight by eliminating pipeline development and deployment cycles
• Maintain governance and lineage throughout AI-enhanced workflows
• Reduce infrastructure costs by consolidating AI execution within your existing lakehouse
• Enable business users to leverage AI capabilities without specialized ML expertise

Simplify analytics by removing ETL and ML dependencies

Data teams spend countless hours building and maintaining ETL pipelines to prepare unstructured data for analysis. AI Functions eliminate this burden through Dremio's zero-copy architecture, which federates queries across data sources and processes unstructured content in place. No more data movement, no more pipeline maintenance, no more delays waiting for engineering resources.

By removing ETL and ML dependencies, organizations reduce operational complexity while accelerating AI adoption. Platform teams can focus on strategic priorities instead of pipeline maintenance, while business professionals and AI agents get immediate access to insights from all data sources—structured and unstructured alike.

Enterprise value:

• Eliminate costly ETL pipelines and reduce data duplication across systems
• Free platform teams from maintenance work to focus on innovation
• Enable federated queries across all sources with consistent governance
• Lower total cost of ownership through autonomous lakehouse operations

Accelerate insight generation with AI-powered SQL functions

Speed matters in today's competitive landscape. AI Functions deliver insights in minutes instead of weeks by combining LLM capabilities with SQL execution. Business teams can ask questions in natural language, while AI agents leverage the AI Semantic Layer to find relevant data and generate accurate answers—all backed by unified, governed data.

This acceleration doesn't compromise accuracy. The AI Semantic Layer provides the business context AI needs to interpret data correctly, ensuring insights are trustworthy and actionable. Whether analyzing sentiment in customer reviews or extracting structured data from PDFs, AI Functions deliver fast, accurate results without manual interpretation or validation.

Enterprise value:

• Transform weeks of development into minutes of SQL authoring
• Empower business professionals to analyze unstructured data independently
• Ensure accuracy through semantic consistency and business context
• Scale AI adoption across the organization with governed, self-service analytics

Automate data classification and summarization at scale

Manual data classification and summarization don't scale. AI Functions enable organizations to process millions of documents, classify sentiment across thousands of customer interactions, and extract structured insights from unstructured content—all at lakehouse scale with minimal operational overhead.

Dremio's purpose-built AI Functions (AI_GENERATE, AI_COMPLETE, AI_CLASSIFY) handle the full spectrum of unstructured data processing needs. From complex multi-field extraction to intelligent summarization and rapid categorization, these functions deliver enterprise-grade capabilities through simple SQL syntax.

Enterprise value:

• Process unstructured data at lakehouse scale without performance degradation
• Automate repetitive classification and summarization tasks
• Maintain consistent quality through governed AI execution
• Reduce manual effort while improving accuracy and coverage

Reduce operational overhead through native AI execution

Operating separate AI infrastructure creates significant overhead: managing compute resources, monitoring model performance, maintaining integrations, and handling security across multiple systems. AI Functions eliminate this burden by executing AI workloads natively within Dremio's autonomous lakehouse.

With autonomous scaling, automatic optimization through Reflections, and consumption-based pricing, organizations get the performance they need at the lowest cost—without manual tuning or capacity planning. Platform teams reduce operational complexity while business teams get faster insights.

Enterprise value:

• Eliminate separate AI infrastructure and associated maintenance
• Benefit from autonomous scaling and optimization
• Pay only for consumption with industry-leading price-performance
• Reduce platform team workload through automated lakehouse operations

Dremio AI Functions are built for enterprise agentic data analysis

Dremio introduces four purpose-built AI Functions designed to address your most pressing agentic analytics needs:

1. AI_GENERATE: Your swiss army knife for complex data extraction

The cornerstone of our AI Functions suite, AI_GENERATE provides flexible, general-purpose processing of unstructured data. This function excels at complex extraction tasks requiring multiple structured fields from source files, enabling you to transform PDFs, documents, and images into queryable data with unprecedented ease. Whether extracting customer details from support tickets or parsing structured information from research papers, AI_GENERATE delivers accurate results through simple SQL syntax.

2. AI_COMPLETE: Intelligent text generation and summarization

Specialized for creative text generation and intelligent summarization, AI_COMPLETE returns VARCHAR outputs perfect for generating executive summaries, creating narrative descriptions, or producing contextual explanations of data patterns. AI agents and business professionals can leverage AI_COMPLETE to synthesize insights across multiple documents, generate reports, or create human-readable interpretations of complex data—all within governed SQL workflows.

3. AI_CLASSIFY: Streamlined categorization at scale

Purpose-built for sentiment analysis and data categorization, AI_CLASSIFY enables rapid classification of text data or unstructured content, returning structured VARCHAR results that integrate seamlessly into your analytical workflows. Process customer feedback, categorize support tickets, or classify documents at lakehouse scale—with consistent accuracy and minimal operational overhead.

4. LIST_FILES: Support for unstructured data processing

To fully unlock the potential of AI insights, Dremio 26.1 introduces essential supporting functionality with the LIST_FILES function. This capability recursively lists files from source directories, enabling batch processing of documents stored in your data lake. Combined with AI Functions, LIST_FILES enables organizations to process thousands of documents with a single query—transforming weeks of pipeline development into minutes of SQL authoring.

Real-world application: Transforming documents into AI insights

Let's explore how an organization with thousands of call log PDFs can leverage AI Functions for instant data load and analysis.:

This single query replaces what would traditionally require:

• A document processing pipeline
• OCR capabilities for scanned documents
• Natural language processing models
• ETL jobs to structure the output
• Ongoing pipeline maintenance

With output schema definition using the WITH SCHEMA clause, the result is a fully structured row that can be used downstream in joins, CTAS, or any other SQL operation—all within a unified, governed environment that maintains lineage and access controls throughout the workflow.

Get agent analytics that accelerate business value

AI Functions deliver immediate benefits that align with Dremio's core value proposition: the fastest path to AI at the lowest cost, without operational burden.

• Fastest path to AI insights: Transform weeks of data pipeline development into minutes of SQL authoring. Business analysts and AI agents can now directly query unstructured data without waiting for engineering resources, accelerating AI adoption across the organization.
• Lowest cost through autonomous operations: Eliminate the need for separate AI infrastructure, specialized tools, and complex pipelines. AI Functions leverage your existing Dremio investment while autonomous lakehouse operations reduce compute consumption and operational overhead—delivering 20× performance at the lowest cost.
• Enhanced data democratization with governance: Enable SQL-savvy analysts to harness AI capabilities without requiring machine learning expertise. The AI Semantic Layer provides business context, while fine-grained access controls ensure security—broadening access to advanced analytics while maintaining enterprise governance.
• Trusted insights through unified data: Process sensitive documents within your secure data lakehouse environment, maintaining lineage and access controls throughout AI-enhanced workflows. Unified data and consistent governance mean AI agents deliver insights teams can trust—without data silos or inconsistent interpretations.

Experience the Agentic Lakehouse advantage

Dremio is the Agentic Lakehouse—the only data platform built for agents and managed by agents. AI Functions are available now, ready to transform how your organization implements and scales AI, whether you're looking to:

• Analyze customer feedback at scale with governed AI execution
• Extract insights from financial documents without data duplication
• Process research papers and technical documentation at lakehouse scale
• Synthesize information from diverse content sources with unified context

Organizations accelerate AI and analytics with unified, governed, and contextual data, while autonomous lakehouse operations reduce cost and eliminate manual tuning. Built on open standards including Apache Iceberg, Polaris, and Arrow, Dremio delivers industry-leading price-performance without pipelines or lock-in.

Book a demo today to see how Dremio AI Functions and the Agentic Lakehouse provide the capabilities you need to unlock the full value of your data and accelerate your path to AI.