OpenAI launches Data agent for ChatGPT Work analytics

OpenAI is positioning Data agent as a way to remove the usual technical barriers to internal analytics. Instead of writing queries or learning a new analytics tool, employees describe what they want in chat, and the feature connects to sources such as Amazon Redshift, BigQuery, Snowflake, and Databricks, along with files in Google Drive and SharePoint.

To interpret data correctly, it draws on each company's own business terms, metric definitions, custom formulas, and relationships between data, pulling those from systems like dbt, Snowflake Horizon, Databricks Genie Ontology, and BI dashboards. It also suggests next steps and people to involve, shares results through Slack or email, and carries out approved tasks through connected tools — with dashboards buildable inside Omni, Oracle BI, Power BI, Sigma, Tableau, and ThoughtSpot.

The rollout is gated by administrators, who decide on deployment and whether to enable the data-source plugin, so the pace of adoption appears to hinge on internal IT choices rather than end users alone. OpenAI says it is already using the feature internally, and NTT Data and others are using it in an alpha program, with non-engineers reportedly creating and updating dashboards in natural language.