OpenAI shows ChatGPT Work for data science

This is a practical how-to release rather than a product launch. The company is addressing a specific audience—data science teams—who often juggle inputs like metric definitions, experiment notes, and business context before producing a polished memo or dashboard specification. The guidance turns that scattered material into a structured first draft that still requires human validation.

The Databricks integration and related tools for data teams suggest OpenAI is framing ChatGPT Work as part of a workflow, not as a replacement for analysts. The reference to the former Codex app shows the features were relocated, not newly invented. The emphasis on review questions and caveats implies the output is meant to be checked, not taken at face value.

What matters for a business reader is the division of labor: the AI accelerates drafting, but the team still owns verification and sharing. The outcome likely hinges on whether these templates hold up on real internal data and whether analysts trust the drafts enough to speed up their review cycles.