Agentic Systems and Databricks Genie: The Future of Autonomous Data Processing

Suteja Kanuri
3 min readFeb 9, 2025

--

Introduction: What Are Agentic Systems?

Agentic systems represent a new wave of AI-driven automation, where Large Language Models (LLMs) serve as the reasoning engine, capable of autonomously deciding actions, executing tasks, and leveraging external tools. Unlike traditional rule-based automation, these systems dynamically adjust their behavior based on inputs and context.

A prime example of an agentic system in data analytics is Databricks Genie, which utilizes LLMs to simplify querying, optimize SQL execution, and enhance decision-making within a data lakehouse environment.

What is the LLM in Agentic Systems?

LLMs form the core intelligence behind agentic systems, acting as the “brain” that processes input, determines intent, and executes appropriate actions. In these systems, the LLM plays several crucial roles:

  1. Understanding and Interpreting User Intent
  • Converts natural language queries into structured commands.
  • Example: “Show me last month’s sales performance” translates into an SQL query.

2. Decision-Making and Planning

  • Determines multi-step execution plans.
  • Example: Genie might decide to pull raw data, apply transformations, and generate a visualization.

3. Tool Use and Function Calling

  • Calls external tools such as SQL functions, APIs, and ML models to retrieve relevant data.
  • Example: Genie queries the metastore for schema details before executing a query.

4. Memory and Context Awareness

  • Retains previous queries and interactions to provide more refined responses.
  • Example: Genie remembers past filters applied by a user.

5. Autonomous Execution and Adaptation

  • Adjusts queries or execution strategies based on feedback or errors.
  • Example: Rewrites an inefficient SQL query for better performance.

LLM as the Brain, Tools as the Hands

A simple framework for understanding agentic systems:

  • LLM = reasoning engine (brain): Processes input, determines intent, and plans actions.
  • External tools (SQL, APIs, ML models) = execution agents (hands): Carry out the actions determined by the LLM.
  • Agentic systems combine both to automate complex workflows, reducing the need for manual intervention and improving efficiency.

Which LLM Powers Databricks Genie?

Databricks Genie is powered by Azure OpenAI models, enabling natural language processing and SQL generation. While Databricks has developed DBRX, an open-source LLM, Genie currently operates using Azure OpenAI’s capabilities.

To ensure data privacy, Databricks has opted into Microsoft’s exemption program, meaning user queries and responses are not stored or reviewed by OpenAI. This ensures enterprise-grade confidentiality and compliance.

How Databricks Genie is an Agentic System

Databricks Genie exemplifies an agentic system by combining LLM-driven reasoning with tool-based execution. Here’s how:

  • Decision-making: Genie autonomously determines the best way to retrieve insights.
  • SQL Functions as Tools: It executes SQL, optimizes queries, and generates dashboards.
  • Trusted Assets: Genie leverages existing tables, ML models, and dashboards to provide accurate insights.
  • Autonomous Execution: Unlike simple chatbots, Genie doesn’t just answer questions — it takes action to optimize workflows.

The Future of Agentic Systems in AI & Data

Agentic systems like Databricks Genie are just the beginning. The next evolution will include:

  • Enhanced Multimodal Agents: Combining text, images, and structured data for richer interactions.
  • Improved Orchestration: AI-driven workflow automation beyond SQL query generation.
  • Tighter Governance and Trust Mechanisms: Ensuring compliance and ethical AI use in enterprise environments.

As LLMs become more advanced, agentic systems will play a key role in democratizing data analytics, making AI-driven decision-making more accessible to both technical and non-technical users.

Conclusion

Databricks Genie represents a powerful shift in how we interact with data. By leveraging LLMs as reasoning engines and external tools as execution agents, Genie embodies the future of autonomous, intelligent data workflows. As agentic systems continue to evolve, they will redefine efficiency, usability, and intelligence in enterprise AI solutions.

--

--

Suteja Kanuri
Suteja Kanuri

No responses yet