OpenAI launches Data agent in ChatGPT Work to analyze company data

OpenAI on September 10, 2026 introduced the Data agent for ChatGPT Work, a tool designed to let employees query and visualize company data through natural-language conversation. According to OpenAI, the agent can investigate metric changes, create interactive dashboards, and surface evidence for findings without requiring users to write SQL or learn a new analytics platform.

Capabilities and integrations

The Data agent connects to a range of approved data sources, including Amazon Redshift, Datadog, Google BigQuery, ClickHouse, Databricks, MongoDB, Snowflake, and common file stores such as Google Drive and SharePoint. OpenAI says the agent uses organization-specific context—business terms, metric definitions, custom calculations and data relationships—sourced from semantic layers and tools like Databricks Genie Ontology, dbt, GitHub, Snowflake Horizon and BI dashboards.

The agent can convert analysis into dashboards with built-in visualizations and supports publishing or interacting with dashboards in Omni, Oracle BI, Power BI, Sigma, Tableau and ThoughtSpot. It can also share findings via Slack or email and execute approved actions through connected tools, OpenAI stated.

Governance, partners and early use

Enterprise administrators determine which data connections and roles are available, and queries respect the connected account’s permissions, including table, row and column restrictions. OpenAI also reported internal use: nearly all of its product team and more than two-thirds of its GTM organization use data agents in ChatGPT Work, enabled by shared business definitions and access rules.

Several platform and BI partners provided statements about the integration: Naresh Chainani, Director of Engineering at AWS, highlighted Redshift connectivity; Ken Wong, Senior Director at Databricks, referenced Genie as a source of context; and Umesh Unnikrishnan, Head of Developer Experiences at Snowflake, emphasized governed access to enterprise data. Customer organizations in the Alpha program—NTT DATA, Thermo Fisher Scientific, ServiceTitan, Zipline and others—reported using the agent to analyze sales, spending and operational metrics, attributed in company quotes.

The Data agent appears as Data in the ChatGPT Work Plugins directory. Administrators can enable it via Workspace settings > Plugins, configure data-source plugins such as Databricks and Snowflake, and manage user access. Example prompts shared by OpenAI include diagnosing metric changes, designing KPI frameworks and creating leadership readouts.


Original source: OpenAI News

Leave a Comment