OpenAI is introducing a Data Agent in ChatGPT Work. The agent connects to approved corporate data sources, analyzes changes in figures, and builds interactive dashboards. Users can direct the analysis in plain language, without writing queries or learning a separate analytics tool.
Anyone within an organization who wants to know why revenue has declined or where spending is increasing usually has to wait for a report or ask someone else to run the analysis. With the new Data Agent, OpenAI aims to eliminate that intermediate step.
The agent runs as a plugin in ChatGPT Work and can be found under the name “Data” in the Plugins directory. Administrators can make it available via Workspace settings, enable the corresponding data source plugins, and set who can use it. Then, simply start a conversation with @Data.
Integrations with data warehouses and BI tools
The list of supported sources is extensive: Amazon Redshift, Datadog, Google BigQuery, ClickHouse, Databricks, MongoDB, and Snowflake. You can also include files from Google Drive and SharePoint in the analysis. For interpretation, the agent relies on semantic layers and definitions from sources such as Databricks Genie Ontology, dbt, GitHub, and Snowflake Horizon.
In addition, the Data agent can build and edit dashboards in Omni, Oracle BI, Power BI, Sigma, Tableau, and ThoughtSpot.
Permissions remain with the source
Administrators determine which data connections are available and which roles can use them. Queries respect the existing permissions of the linked account, including restrictions at the table, row, and column levels.
After an analysis, ChatGPT Work can suggest next steps, indicate who should be involved, and share findings via Slack or email. The agent executes approved actions using connected tools.
Tip: Tableau is entering the next phase of analytics with AI agents