Genie Code, a new AI agent from Databricks, aims to automate complex data engineering and analytics tasks, marking the company’s latest move to expand agent-driven development for enterprise data teams.

The Genie Code AI agent can perform multi-step workflows such as building data pipelines, debugging failures, creating dashboards and maintaining production systems. Databricks said the system more than doubled the success rate of leading coding agents in real-world data science tasks, rising from 32.1 per cent to 77.1 per cent.

The launch also coincides with Databricks’ acquisition of Quotient AI, a startup specialising in evaluation and reinforcement learning for AI agents, to strengthen quality monitoring and continuous improvement of its AI systems.

Agentic Data Work for Enterprise Data Teams

Databricks said Genie Code is designed to shift data work from AI-assisted coding to agent-driven workflows, where AI plans, executes and maintains data systems while humans retain oversight.

“Software development has shifted from code-assistance to full agentic engineering in the past six months,” said Ali Ghodsi, co-founder and chief executive officer of Databricks. “Genie Code brings this revolution to data teams. We’re moving from a world where data professionals are assisted by AI to one where AI agents do the work, guided by humans.”

The company describes this shift as “Agentic Data Work”, where autonomous systems handle operational tasks across the data lifecycle.

Genie Code builds on Genie, Databricks’ conversational data interface that allows employees to query enterprise data using natural language. The new capability extends the platform to data engineers, analysts and data scientists by automating technical work required to move projects from idea to production.

Enterprise Context and Data Governance

According to Databricks, existing coding agents often struggle with data-centric workloads because they lack access to business context such as data lineage, governance policies and usage patterns.

Genie Code integrates with Unity Catalog, Databricks’ governance layer, allowing the AI agent to enforce access controls, understand enterprise data relationships and comply with audit requirements.

The company said the agent can perform a range of advanced tasks, including managing machine learning workflows, deploying models and maintaining data pipelines.

It also monitors Lakeflow pipelines and AI models for failures or anomalies, and can automatically adjust resources or investigate issues before human intervention.

“At SiriusXM, Genie Code supports everything from authoring notebooks and complex SQL to reasoning through table relationships and debugging pipelines,” said Bernie Graham, vice president of data engineering at SiriusXM. “It acts as a hands-on development partner that helps our data teams deliver high-quality work in less time.”

Similarly, Repsol said the technology can accelerate development workflows without sacrificing governance.

“Instead of stitching together notebooks, pipelines and models manually, we can hand off complex workflows to an AI partner that understands our data, governance and business context,” said Emilio Martín Gallardo, principal data scientist at Repsol.

Continuous Evaluation Through Quotient AI Acquisition

To improve the reliability of agent-driven systems, Databricks also announced the acquisition of Quotient AI, a company focused on evaluating and improving AI coding agents.

Quotient’s technology automatically monitors agent performance, measuring answer quality, detecting regressions and identifying failures. The system feeds results into a reinforcement learning loop to improve AI behaviour over time.

The startup’s founders previously worked on quality improvements for GitHub Copilot, according to Databricks.

The company said embedding Quotient’s capabilities into Genie and Genie Code will allow AI systems to continuously evaluate and improve their performance in production environments.