OpenAI has introduced a Data agent in ChatGPT Work that is meant to turn ordinary business questions into analysis and shareable dashboards. Instead of waiting for a data team to build a report, an employee can ask why sales changed, where spending is rising, or which accounts need attention, then continue the investigation in the same conversation.
From connected data to a governed answer
The Data agent connects to approved sources including Amazon Redshift, Datadog, Google BigQuery, ClickHouse, Databricks, MongoDB, and Snowflake. It can also bring files and documents from Google Drive and SharePoint into an analysis. OpenAI says the agent uses an organization’s metric definitions, custom calculations, business terms, and data relationships from semantic layers and trusted sources such as dbt, GitHub, Snowflake Horizon, Databricks Genie Ontology, and BI dashboards.
That context is important because a plain-language question is not a metric definition. “Revenue” can mean different things across a company, and a dashboard can be technically correct while answering the wrong business question. OpenAI’s design puts the organization’s existing semantics in the retrieval path rather than asking the model to infer every definition from raw tables.
Dashboards without a separate BI project
After investigating a result, the Data agent can produce an interactive dashboard with built-in visualizations. Teams can edit, share, and refresh those dashboards, and OpenAI says they can also build or interact with views in Omni, Oracle BI, Power BI, Sigma, Tableau, and ThoughtSpot. The product page frames the workflow as conversational: ask a question, inspect the evidence, refine the analysis, and turn it into a view others can use.
The agent can also recommend next steps, identify who should be involved, share findings through Slack or email, and carry out approved actions through connected tools. That makes the feature more than a natural-language query layer, but it also raises the stakes for permissions and confirmation. A wrong chart is misleading; an unreviewed action can change a real workflow.
Existing permissions stay in the loop
OpenAI says enterprise administrators choose which data connections are available and which roles can use them. Queries enforce the connected account’s existing table, row, and column restrictions. In effect, the agent is intended to inherit the organization’s access model rather than create a parallel data universe that exposes every table to every employee.
OpenAI also says nearly all of its product team and more than two-thirds of its go-to-market organization use data agents internally, while organizations in its alpha program are applying the tool to sales, spending, reporting errors, supply-base analysis, and operations. These are company-reported examples, not independent evaluations of accuracy or return on investment.
Newsroom analysis
The product’s larger bet is that the bottleneck in business analytics is often not charting but access to shared definitions and permissioned context. If the Data agent can respect those boundaries while showing evidence, it could move more exploratory analysis into the hands of operators. The trade-off is that organizations will need stronger semantic governance, clear approval flows, and a habit of checking the underlying data before a conversational answer becomes a decision.