Genie Code transforms ideas in data engineering, data science, and analytics into autonomous production systems.
Data and AI company Databricks has launched Genie Code, an autonomous AI agent that fundamentally changes the way we work with data. Genie Code can perform complex tasks such as building pipelines, debugging failures, shipping dashboards, and maintaining production systems. Databricks has found that Genie Code more than doubles the success rate of leading coding agents on real-world data science tasks. Just as agent coding tools have transformed software engineering, moving developers from autocomplete-style assistance to agent-driven development, Genie Code brings a similar paradigm shift to data engineering, data science, and analytics.
Genie Code is a new addition to Genie that allows knowledge workers to chat with data and get instant, trusted answers using context and semantics captured by Unity Catalog. Genie Code extends this approach to data professionals, handling the complex engineering required to go from idea to production across all enterprise data. Additionally, today, Databricks announced the acquisition of Quotient AI, an innovator in AI agent evaluation and reinforcement learning, to bring continuous evaluation directly into Genie and Genie Code.
Marketing Technology News: MarTech Interview @ TrafficGuard CPO, Miguel Lopes
The rise of agentic data work
Today’s data tools treat AI as a helper, writing code, running local tests, and iterating over it. This leaves data teams with the heavy lifting of planning, orchestration, operations, validation, and maintenance. Genie Code reverses this approach. Reason through problems, plan multi-step approaches, write and validate production-grade code, and maintain results, all while putting humans in control of important decisions.
“In the last six months, software development has moved from code-assisted to full agent engineering,” said Ali Ghodsi, co-founder and CEO of Databricks. “Genie Code brings this revolution to data teams. We are moving from a world where data professionals are supported by AI to a world where AI agents do their work with human guidance. We call this agential data work. This will fundamentally change the way enterprises make decisions.”
Genie Code Features
Existing agent coding tools have difficulty performing data tasks because they lack access to important context such as lineage, usage patterns, and business semantics. Genie Code helps teams bridge the context gap to ensure the high level of accuracy and governance required for production environments. Genie code:
- Working as an expert machine learning engineer: Genie Code handles complete ML workflows end-to-end. Infer complex problems and plan, create, and deploy models while recording experiments in MLflow and fine-tuning service delivery endpoints for best performance.
- Incorporate deep data engineering expertise: Novice engineers may write scripts to manipulate test data, but Genie Code designs like an advanced architect. Build change data capture workflows that account for differences between staging and production environments, and enforce data quality expectations.
- Proactive maintenance and optimization: Genie Code monitors Lakeflow pipelines and AI models in the background, triages failures, and investigates anomalies. Autonomously analyzes agent tracking to correct hallucinations and adjust resource allocation before human intervention.
- Understand the corporate context: Integrated with Unity Catalog, Genie Code enforces existing governance policies and access controls. Understand business semantics and audit requirements and integrate corporate data, including data from external platforms.
- improves over time: The Genie Code gets smarter the more your team uses it. Through persistent memory, it automatically updates internal instructions based on past interactions and coding settings. Databricks found that Genie Code has more than twice the success rate of leading coding agents (32.1% to 77.1%) on real-world data science tasks.
Marketing Technology News: Disrupt or be disrupted: AI’s wake-up call for B2B marketers
“At SiriusXM, Genie Code supports everything from writing notebooks and complex SQL to inferring with table relationships and debugging pipelines,” said Bernie Graham, VP of Data Engineering at SiriusXM. “It serves as a hands-on development partner that helps data teams deliver high-quality work in less time.”
“Genie Code changes the way our data team operates,” said Emilio Martín Gallardo, Principal Data Scientist for Data Management and Analytics at Repsol. “Instead of manually piecing together notebooks, pipelines, and models, we can hand off complex workflows to AI partners who understand our data, governance, business context, and internal libraries like our Repsol artificial intelligence products. This accelerates everything from time-series forecasting to production deployment without sacrificing rigor or control.”
Strengthen continuous evaluation by acquiring commercial AI
To close the loop on production quality, Databricks acquired Quotient AI. Quotient automatically monitors agent performance, measures answer quality, detects regressions early, pinpoints failures, and feeds reinforcement learning loops to ensure agents continue to improve over time. Quotient’s founders previously led quality improvements at GitHub Copilot and bring deep expertise in evaluating AI coding systems. By building these capabilities into Genie Code, Databricks ensures that your data and AI systems not only run in production, but are continuously improved.
