Glonce

OpenAI introduces Data agent in ChatGPT Work

agentsThe Glonce editorsOpenAI🔥73Glonce rating73Impact70Novelty65Relevance90Surprise45Players95Credibility90
OpenAI introduces Data agent in ChatGPT Work

OpenAI has introduced a new Data agent inside ChatGPT Work, its enterprise chat product. The company frames the need with everyday business questions that normally require waiting for a report or another person to run the numbers: why sales slowed, where spending is rising, and which issues threaten renewals in a company's largest accounts. The Data agent is built to let people answer such questions themselves: it connects to a company's data, investigates what changed, and builds interactive dashboards that can be shared, all from a plain-language question rather than a written query or a separate analytics tool. Users can ask follow-up questions to dig into the results and review the evidence behind each finding.

The agent connects to approved data sources including Amazon Redshift, Datadog, Google BigQuery, ClickHouse, Databricks, MongoDB and Snowflake, among others, and can also pull files and documents from Google Drive and SharePoint into an analysis. To interpret the data correctly, it uses an organization's own business terms, metric definitions, custom calculations and data relationships, drawn from semantic layers and sources such as Databricks Genie Ontology, dbt, GitHub, Snowflake Horizon and existing BI dashboards.

Enterprise administrators decide which data connections are available and which roles may use them. Every query enforces the connected account's existing permissions, including table, row and column restrictions.

The Data agent turns an analysis into an interactive dashboard with built-in visualizations that a team can edit, share and refresh, styled to an organization's own brand guidelines if desired. It can also build and interact with dashboards directly inside Omni, Oracle BI, Power BI, Sigma, Tableau and ThoughtSpot, directed entirely in plain language. Beyond analysis, it can recommend next steps, identify who else needs to be involved, share findings through Slack or email, and carry out the actions a user approves through connected tools.

OpenAI says it already relies on the same capabilities broadly inside its own company: nearly all of its product team and over two-thirds of its go-to-market organization use data agents in ChatGPT Work to analyze company data themselves. The company credits this to its data team building shared business definitions, setting access rules, and putting safeguards in place for sensitive data. Outside OpenAI, organizations in its Alpha program, including NTT Data, Thermo Fisher and ServicePiston, are already using the Data agent in ChatGPT Work to analyze sales and spending, catch reporting errors, and decide which opportunities to pursue and how to staff them.

The Data agent is listed as Data in the Plugins directory inside ChatGPT Work. An administrator can make it available or install it for a team through Workspace settings > Plugins, then enable and configure the relevant data-source plugins, such as Databricks or Snowflake, and control who can use them. Once installed, a user completes any required account-connection steps and starts a conversation with @Data, for instance asking it to diagnose why weekly active users changed last week, identify likely drivers, compare the results against prior periods, and recommend the next checks.

Key facts

  • OpenAI's new Data agent in ChatGPT Work investigates a company's data and builds shareable interactive dashboards from plain-language questions, without anyone writing a query.
  • It connects to sources including Amazon Redshift, Google BigQuery, Snowflake, Databricks, ClickHouse and MongoDB, plus files from Google Drive and SharePoint.
  • Enterprise administrators control which data connections and roles are enabled, and every query enforces the connected account's existing table, row and column permissions.
  • OpenAI says nearly all of its product team and over two-thirds of its go-to-market organization already use data agents in ChatGPT Work to analyze their own data.
  • Alpha-program customers, including NTT Data, Thermo Fisher and ServicePiston, use the Data agent to analyze sales and spending, catch reporting errors, and decide which opportunities to pursue.

Why it matters

Getting a data-backed answer inside a company usually means waiting for an analyst or a scheduled report. OpenAI is positioning the Data agent as a way to remove that wait by pairing a chat interface with an organization's own metric definitions and data relationships, so a person can ask a business question directly and get an investigated, shareable dashboard back instead of a request in a queue. OpenAI reports heavy internal adoption of the same approach among its own product and go-to-market staff, which it credits to work by its data team: creating shared business definitions, setting access rules and putting safeguards in place for sensitive data.

Who it affects

Anyone who needs a data-backed answer without waiting on an analyst, especially people in sales, finance and account-management roles who can now ask business questions directly, and the enterprise administrators who choose which data sources and user roles are enabled. It targets ChatGPT Work customers whose data already sits in the supported warehouses and BI tools; among the Alpha-program organizations already using it, OpenAI names NTT Data, Thermo Fisher and ServicePiston, which use the Data agent to analyze sales and spending, catch reporting errors, and decide which opportunities to pursue and how to staff them.

How to use it

The Data agent is listed as Data in the Plugins directory inside ChatGPT Work. An administrator makes it available or installs it for a team through Workspace settings > Plugins, then enables and configures the relevant data-source plugins, such as Databricks or Snowflake, and sets who can use them. A user completes any required account-connection steps and starts a conversation with @Data, asking a business question in plain language, for example asking it to diagnose why weekly active users changed, identify likely drivers, compare against prior periods, and recommend the next checks. The result can be a shareable, brand-styled dashboard inside ChatGPT Work, or a dashboard built directly in Omni, Oracle BI, Power BI, Sigma, Tableau or ThoughtSpot. OpenAI states no pricing or subscription tier for the Data agent or for ChatGPT Work generally in this announcement.

How solid is it

The figures and case studies come from OpenAI's own announcement, not independent testing. The internal-adoption numbers, nearly all of the product team and over two-thirds of the go-to-market organization, are self-reported, as are the three named Alpha-program customers and their use cases. OpenAI attributes its internal results to deliberate preparation rather than to the agent alone, so how well those numbers would transfer to a customer that has not done the same groundwork is untested here.

Risks and caveats

OpenAI gives no release date, rollout timeline or region availability for the Data agent, and does not clarify whether it is already generally available to ChatGPT Work customers, despite giving general install instructions through the Plugins directory, or still limited to Alpha program participants like the three named organizations. No underlying AI model is named, and no individual, whether an executive, an engineer or a customer contact, is named behind the announcement or the case studies. Because the agent's data access mirrors the connected account's existing permissions, how well it performs in practice will depend on how carefully an organization has already set up its own roles, definitions and restrictions.