Overview
On September 10, 2026 OpenAI put a data analysis agent inside ChatGPT Work, invoked as `@Data` from the plugin directory. It targets everyday business questions: why sales slowed, where spending is rising, which large accounts are at risk of not renewing. OpenAI's framing is that answers to those questions normally require waiting on a report or asking someone else to run the analysis, and the Data agent compresses that into a single conversation.
Key Features
- Direct data connections: Connects to approved sources including Amazon Redshift, Datadog, Google BigQuery, ClickHouse, Databricks, MongoDB and Snowflake, and can pull files and documents from Google Drive and SharePoint into an analysis.
- Interprets data through company definitions: Beyond joining tables, it reads an organisation's own business terms, metric definitions, custom calculations and data relationships. That context comes from semantic layers and trusted sources such as Databricks Genie Ontology, dbt, GitHub, Snowflake Horizon and existing BI dashboards.
- Permissions carry into queries: Enterprise administrators choose which data connections are available and which roles can use them. Queries enforce the connected account's existing permissions, including table-, row- and column-level restrictions.
- Dashboards from plain language: An analysis can be turned into an interactive dashboard with built-in visualisations that teammates can edit, share and refresh. Brand guidelines can be supplied so output matches company look and feel.
- Works with existing BI tools: Beyond its own dashboards, it can build and interact with dashboards in Omni, Oracle BI, Power BI, Sigma, Tableau and ThoughtSpot.
- From analysis to action: ChatGPT Work can recommend next steps, identify who needs to be involved, share findings through Slack or email, and carry out user-approved actions through connected tools.
Use Cases
- Tracking a product end to end from install and activation through retention, to find where users get stuck
- Consolidating company performance metrics scattered across multiple systems into one overview dashboard
- Investigating reporting errors and checking metric definitions and calculation logic
- Letting sales and marketing teams update their own dashboards instead of queueing for the data team
Pros
- Ties natural-language querying to the existing permission system rather than a parallel one
- Output is an interactive dashboard rather than a one-off export
- Reads company metric definitions, so answers match internal phrasing
- Aimed at non-engineering roles without SQL or BI training
Pricing
Delivered with ChatGPT Work enterprise subscriptions, with pricing set through OpenAI's commercial terms. Which data connections are available is configured by enterprise administrators.
Summary
The Data agent fills the querying gap. Getting from data to decisions usually bottlenecks on people rather than compute — the number of staff who can write SQL or drive a BI tool is always small. Binding natural-language querying to the permission system and turning the result directly into a shareable dashboard is especially useful for non-engineering roles in mid-size and large organisations. OpenAI says almost its whole product team and more than two-thirds of its go-to-market organisation already use it, and Alpha customers such as NTT Data and Thermo Fisher have it in live business analysis. Data governance determines how far it spreads: the clearer the admin-configurable connections and roles, the less friction adoption meets.
Version History
- Data agent launch (2026-09-10): Released in the ChatGPT Work plugin directory with support for Redshift, BigQuery, ClickHouse, Databricks, MongoDB, Snowflake and more