The hunt for the mythical all-knowing data scientist may finally be over — if a new AI-powered build tool delivers on its promise.
Dataiku has launched Cobuild, an AI agent that builds data and analytics projects from plain-language descriptions, integrating directly with Snowflake’s data platform to keep every workflow inside an organization’s existing governance structure.
The tool is designed to democratise AI development without sacrificing the oversight that enterprise IT and compliance teams require. A user describes what they want — a churn model weighted by contract size, a data quality check across four quarters of account activity — and Cobuild assembles the project as a visual, step-by-step Dataiku workflow. Every node is inspectable and editable. Nothing reaches production without human sign-off.
That last point is the core of Dataiku’s pitch. Where conventional AI coding tools produce blocks of code that only a developer can audit, Cobuild produces a visual flow that the person closest to the business problem can evaluate and approve directly. The output can include data preparation steps, joins, custom-trained models, charts for interpreting model bias, and, where needed, polished web applications that draw on the data and models already created.
When running on Snowflake, all model calls — including those to Anthropic’s Claude and OpenAI’s GPT — are routed through Snowflake Cortex within the customer’s account. Data, conversations, and computation stay within Snowflake’s environment, meaning the governance model organizations have already built around their Snowflake investment applies automatically to everything Cobuild generates.
Also Read: Google I/O 2026 Day 1: The Agentic Era Is Here
More than 350 organizations currently run Dataiku on Snowflake. For those teams, Cobuild expands the pool of people who can contribute to AI development without requiring a data scientist’s involvement at every step — freeing specialist teams to focus on work that genuinely demands their expertise.
“Companies are still searching for their golden goose data scientist,” the company said at launch — someone who combines graduate-level mathematics with business acumen, engineering skills, and executive communication. “With Cobuild, we’re here to turn everyone into a golden goose.”
