---
title: "Agentic data engineering with Genie Code and Lakeflow"
description: "Genie Code is presented as a natural-language assistant for developing, orchestrating, deploying, and debugging data pipelines and jobs. It uses pipeline and job context, including code, configuration, run results, Unity Catalog metadata, lineage, popularity, and code samples, to help engineers discover datasets and understand data flows. Engineers can describe pipelines or jobs, and Genie Code can generate Spark Declarative Pipelines with Bronze, Silver, and Gold layers, sources, transformations, data quality expectations, and outputs, then configure orchestration, schedules, dependencies, Auto Loader, AutoCDC flows, and Declarative Automation Bundles. It also analyzes failures and unexpected row-count or schema changes, proposes cross-file updates with reviewable diffs, and supports extensions through custom instructions, agent skills, and MCP servers. The stated result is faster development and guided debugging while workflows remain aligned with Unity Catalog governance, performance, and data quality standards; future plans include background failure response and cluster right-sizing."
---

# Agentic data engineering with Genie Code and Lakeflow

[Databricks](https://yomu.fyi/company/databricks) · Gal Oshri, Camiel Steenstra, Lennart Kats, Joanna Zouhour · Apr 28, 2026

**Type:** Explainer

## Summary

Genie Code is presented as a natural-language assistant for developing, orchestrating, deploying, and debugging data pipelines and jobs. It uses pipeline and job context, including code, configuration, run results, Unity Catalog metadata, lineage, popularity, and code samples, to help engineers discover datasets and understand data flows. Engineers can describe pipelines or jobs, and Genie Code can generate Spark Declarative Pipelines with Bronze, Silver, and Gold layers, sources, transformations, data quality expectations, and outputs, then configure orchestration, schedules, dependencies, Auto Loader, AutoCDC flows, and Declarative Automation Bundles. It also analyzes failures and unexpected row-count or schema changes, proposes cross-file updates with reviewable diffs, and supports extensions through custom instructions, agent skills, and MCP servers. The stated result is faster development and guided debugging while workflows remain aligned with Unity Catalog governance, performance, and data quality standards; future plans include background failure response and cluster right-sizing.

## Context

Data engineers traditionally spend weeks finding data, building transformations, stitching together jobs, and fixing failures. The capabilities described aim to reduce that manual effort while keeping pipelines and jobs aligned with governance, operational, performance, and data quality standards.

## Approach / What changed

Genie Code works from natural-language requests and accesses pipeline and job code, configuration, context, and run results. It searches data assets using Unity Catalog metadata and lineage, generates and modifies Lakeflow Spark Declarative Pipelines and Lakeflow Jobs, works with Declarative Automation Bundles, analyzes failures, shows proposed diffs, and can be extended with custom instructions, agent skills, and MCP servers.

## Takeaways

- Genie Code can generate a Spark Declarative Pipeline with Bronze, Silver, and Gold layers, sources, transformations, data quality expectations, and outputs from a plain-language description.
- For jobs and failures, it can configure tasks, dependencies, and schedules, analyze errors, propose updates across relevant files, and present diffs for review before changes are applied.
- Declarative Automation Bundles let Genie Code add resources, update configurations, validate bundles, and deploy targets without hand-writing YAML, supporting source control, testing, and CI/CD.

**Tags:** [Data Pipelines](https://yomu.fyi/topic/data-pipelines), [Genie](https://yomu.fyi/topic/genie), [Lakeflow](https://yomu.fyi/topic/lakeflow), [MCP](https://yomu.fyi/topic/mcp), [Unity Catalog](https://yomu.fyi/topic/unity-catalog)

- Source: [Databricks](https://www.databricks.com/blog/agentic-data-engineering-genie-code-and-lakeflow)
- Source URL: https://www.databricks.com/blog/agentic-data-engineering-genie-code-and-lakeflow
- Ingested by Yomu: 2026-08-31T03:40:54.656Z

[Read original post](https://www.databricks.com/blog/agentic-data-engineering-genie-code-and-lakeflow)