The new capability enables developers to build, migrate and troubleshoot Dataiku projects through Claude Code, Codex, Cursor and other agentic coding tools while retaining enterprise governance.
Dataiku has introduced Dataiku Headless, a new capability designed to connect agentic coding tools with its governed enterprise data and AI platform. The offering allows developers to build, modify and investigate Dataiku projects from environments such as Claude Code, OpenAI Codex and Cursor, without having to leave their preferred coding workflows.
The launch addresses a growing enterprise challenge: while AI coding agents are accelerating software and data development, the resulting work can remain difficult for wider teams to inspect, govern, maintain and operationalise. Dataiku aims to bridge that gap by turning work initiated through coding agents into structured Dataiku projects that business and technical teams can review and operate.
Headless brings Dataiku Cobuild, the company’s natural-language agentic interface, into developers’ existing tools. It acts as a control layer through which coding agents can inspect and operate Dataiku, delegating project creation and modification to Cobuild when required.
Developers can use the capability to create data pipelines, datasets, transformations, machine learning models, dashboards and custom AI agents. Rather than leaving the output as standalone scripts, the work is represented as visual Dataiku assets that can be inspected and managed within the platform.
“Developers aren’t going to stop working in Claude Code or Codex, and we don’t think they should have to.”
— Jed Dougherty, SVP of AI & Platform, Dataiku
Another significant application is workflow modernisation. Coding agents can analyse legacy processes built in tools such as Excel or Alteryx and instruct Cobuild to recreate them as governed, visual Dataiku projects. The approach is intended to accelerate migration at scale, potentially automating repetitive work across hundreds or thousands of workflows while converting one-time code translations into maintainable pipelines.
Headless also extends to platform administration and troubleshooting. Administrators can ask coding agents to identify failing jobs across multiple projects, investigate underlying issues and help remediate them by changing project or administrative settings.
“Developers aren’t going to stop working in Claude Code or Codex, and we don’t think they should have to,” said Jed Dougherty, SVP of AI & Platform at Dataiku. He said the objective is to convert the speed of AI-assisted development into work that the wider organisation can review, run and use in production.
Dataiku Headless is available as a plugin through the Claude Code and Codex marketplaces. It is also compatible with Cursor, Snowflake CoCo, GitHub Copilot, AWS Kiro and other agentic coding tools.
The introduction reflects a broader shift in enterprise AI development. As coding agents make it easier to generate scripts, models and workflows, organisations must ensure that these outputs do not become disconnected from established governance, operational oversight and collaboration processes. Dataiku’s approach seeks to preserve developer flexibility while ensuring that AI-generated work can move into governed enterprise environments.
