OpenAI’s New Data Agent Lets Businesses Analyze Company Data, Build Dashboards and Find Insights With AI
The new product was announced on September 10, 2026, as part of OpenAI's broader push to turn ChatGPT Work into a platform for completing complex business tasks rather than simply generating text.
The Data agent can connect to approved company data sources, use an organization's business definitions and investigate questions such as why sales changed, where spending is increasing or which accounts may be at risk.
Instead of requiring employees to write SQL queries or wait for a data analyst to prepare a report, the system is designed to let users ask questions directly.
OpenAI says the agent can then investigate the underlying data, explain its findings, create interactive dashboards and allow users to continue refining the analysis in the same conversation.
This makes the launch particularly relevant to businesses that have large amounts of data but want more employees to work with it.
What Is the OpenAI Data Agent?
The Data agent is an AI-powered data analysis system built into ChatGPT Work.
It is designed to connect AI reasoning with the data infrastructure companies already use.
Users can ask questions in everyday language rather than manually writing database queries or learning a specialized analytics platform.
For example, a business user could ask:
Why did our weekly active users fall last month?
The Data agent can investigate the available data, compare periods, identify possible drivers and provide evidence behind its conclusions.
A user can then ask follow-up questions without starting the analysis from scratch.
OpenAI describes this as a conversational way to investigate business data and turn the results into actionable analysis.
Which Data Sources Can the Data Agent Connect To?
One of the most important aspects of the new system is its ability to connect with existing enterprise data platforms.
OpenAI lists support for data sources including:
- Amazon Redshift
- Datadog
- Google BigQuery
- ClickHouse
- Databricks
- MongoDB
- Snowflake
- Google Drive
- SharePoint
The company says additional data sources and integrations are available through its broader plugin ecosystem.
This means companies do not necessarily need to move their data into a new OpenAI-specific database before using the Data agent.
Instead, the agent can work with information that already exists inside approved business systems.
That approach is important for enterprise adoption because companies often have years of data stored across multiple platforms.
The Data Agent Understands Business Context
Simply connecting an AI model to a database does not automatically make it understand what the numbers mean.
Different organizations can use completely different definitions for the same metric.
For example, one company may define an active customer differently from another.
The Data agent is designed to address this problem by using business context, semantic layers, metric definitions, custom calculations and relationships between datasets.
OpenAI says it can use trusted context from systems such as Databricks Genie Ontology, dbt, GitHub, Snowflake Horizon and existing BI dashboards.
This allows the agent to interpret data according to an organization's existing definitions instead of treating every database field as an isolated number.
Why Business Context Matters
Consider a company asking:
"Why did revenue fall last quarter?"
The answer may depend on how revenue is calculated, which customers are included, how refunds are handled and which business segments are being compared.
Without that context, an AI system could produce an answer that looks convincing but uses the wrong definition.
The Data agent's integration with trusted business context is therefore one of the more important parts of the product.
Employees Can Ask Questions Without Writing SQL
Traditional data analysis often requires technical knowledge.
A business employee may know exactly what they want to understand but not know how to query the database.
That creates a bottleneck.
The Data agent attempts to remove that barrier.
Instead of writing a query such as:
SELECT ...
users can describe what they want in normal language.
For example:
"Compare our sales performance across regions over the last six months and identify the biggest changes."
The system can then perform the analysis and explain what it found.
OpenAI says this is intended to make sophisticated data analysis accessible to more people across an organization.
Data Agent Can Investigate Changes, Not Just Display Numbers
Another important distinction is that the Data agent is designed to investigate questions rather than simply generate charts.
A dashboard might tell a manager that sales declined by 8%.
The Data agent can be asked what caused the decline.
It can investigate the available data, compare periods and identify possible drivers.
Users can then continue asking follow-up questions.
For example:
"Which products contributed most to the decline?"
Then:
"Was the decline concentrated in a particular region?"
And finally:
"Create a dashboard showing those trends."
The workflow can remain inside the same conversation.
This conversational approach is one of the main differences between an AI data agent and a traditional static dashboard.
OpenAI Data Agent Can Build Interactive Dashboards
The system is not limited to text responses.
OpenAI says the Data agent can turn analysis into interactive dashboards with built-in visualizations.
Teams can edit, share and refresh those dashboards after they are created.
Users can also provide brand guidelines so dashboards can be tailored to an organization's visual style.
This could reduce the amount of manual work required to turn an analysis into a presentation-ready business report.
Existing BI Tools Can Also Be Used
The Data agent can work with dashboards and BI environments from platforms including:
- Omni
- Oracle BI
- Microsoft Power BI
- Sigma
- Tableau
- ThoughtSpot
OpenAI says users can direct the analysis through natural-language instructions while continuing to work with these existing tools.
This is important because businesses are unlikely to abandon established BI platforms simply because an AI data tool becomes available.
Integration therefore becomes more valuable than replacement.
Enterprise Permissions Still Apply
Giving an AI agent access to company data creates obvious security concerns.
OpenAI says administrators control which data connections are available and which roles can use them.
The Data agent also respects existing permissions, including table-level, row-level and column-level restrictions.
This means a user should not automatically gain access to every piece of information simply because an AI system is connected to a company database.
The permissions configured by the organization continue to determine what information can be accessed.
That is an important requirement for enterprise environments where financial, customer, employee and operational information may be subject to different access policies.
The Data Agent Can Work With Files and Documents
Business analysis is not always limited to structured database tables.
Companies also store important information in documents, spreadsheets and other files.
OpenAI says the Data agent can bring files and documents from services such as Google Drive and SharePoint into analysis.
This expands the potential use cases.
For example, a team could combine structured sales information with internal documents containing business context.
An AI agent could then analyze both sources when answering a question.
That is useful because real-world business decisions rarely depend on one database alone.
OpenAI Is Already Using Data Agents Internally
OpenAI says the technology behind the Data agent is already being used across the company.
According to OpenAI, nearly all of its product team and more than two-thirds of its go-to-market organization use data agents in ChatGPT Work to analyze company data themselves.
The company's data team created shared business definitions, access rules and safeguards to make this possible.
This internal experience appears to have influenced the product.
Rather than treating AI-powered analytics as an experimental feature, OpenAI is positioning it as a tool that can be used throughout an organization.
Companies Are Already Testing the Data Agent
OpenAI says organizations including NTT DATA, Thermo Fisher, ServicePiston, Zipline, Empower, Turing, micro1 and others have tested or used the Data agent through its Alpha program.
The reported use cases cover a wide range of business activities.
Examples include:
- Sales analysis
- Spending analysis
- Operational reporting
- Customer behavior
- AI adoption tracking
- Marketing performance
- Supply-chain analysis
- Dashboard creation
- Business development
This suggests the product is not restricted to one particular department.
Different teams can use the same underlying system for different business questions.
Data Agent Can Help Non-Technical Teams
One of the biggest opportunities for the product is democratizing access to data analysis.
A company may already employ data engineers and analysts, but those teams cannot necessarily answer every question immediately.
If hundreds of employees have questions about data, analysts can quickly become a bottleneck.
The Data agent could allow employees to investigate routine questions themselves while analysts focus on more complicated work.
For example, a sales manager could investigate regional performance without requesting a new report.
A marketing manager could analyze campaign results.
A finance team could examine spending trends.
An operations team could investigate changes in performance.
The data team can then focus on governance, data quality and more advanced analysis.
OpenAI Shows Several Business Use Cases
OpenAI's own examples show how the Data agent can be used across different business functions.
Diagnosing Metric Changes
Users can ask the agent to investigate why an important metric changed.
The agent can compare time periods, identify possible drivers and suggest additional checks.
Building KPI Frameworks
The system can help create KPI structures containing primary metrics, drivers, guardrails, targets and data-validation requirements.
This could help teams move from raw data toward a structured measurement system.
Creating Leadership Reports
The Data agent can turn business metrics into leadership-ready updates containing actual results, comparisons, drivers, caveats and recommended actions.
This could reduce the manual work involved in preparing recurring management reports.
The Product Fits OpenAI's Larger Agent Strategy
The Data agent arrives shortly after OpenAI introduced its Agents API, another September 2026 product aimed at helping developers build long-running AI agents.
However, the two products target different audiences.
The Agents API is developer infrastructure.
The Data agent is an end-user business application inside ChatGPT Work.
Both reflect the same larger trend: OpenAI is moving AI beyond simple question-and-answer interactions toward systems that can investigate problems and perform multi-step work.
This also complements OpenAI's recent expansion into specialized professional workflows, including the financial-services product already covered by TheInfoBytes.
How Data Agent Differs From Traditional BI Software
Traditional business intelligence platforms are extremely powerful, but they often require users to understand dashboards, filters, data models or query languages.
The Data agent adds a conversational interface on top of business data.
Instead of navigating through multiple reports, a user can simply ask a question.
The AI can then investigate the information and generate an explanation.
This does not mean traditional BI platforms become unnecessary.
Businesses still need reliable data pipelines, governance, semantic models, permissions and validated reporting systems.
The AI layer works best when it is connected to those trusted foundations.
In that sense, the Data agent is less about replacing data infrastructure and more about making that infrastructure easier for employees to use.
Security and Governance Will Be Critical
AI-powered access to business data introduces new risks.
An organization needs to know:
- Which data the AI can access
- Which employees can use it
- How permissions are enforced
- How generated findings are validated
- How sensitive information is handled
- Which actions the agent is allowed to perform
OpenAI says administrators control data connections and user availability, while existing data permissions are enforced.
However, businesses should still treat AI-generated analysis as something that requires appropriate review, particularly when decisions involve financial, legal, operational or personnel consequences.
An AI system can accelerate analysis, but organizations remain responsible for how they use its results.
How to Get Started With OpenAI Data Agent
OpenAI says the Data agent is available through the Plugins directory in ChatGPT Work.
Administrators can make the Data plugin available to teams through workspace settings and configure the relevant data-source plugins.
Once the appropriate connections are configured, users can start a conversation with @Data and ask their business question.
The initial setup therefore depends on organizational access and data-source configuration rather than simply opening a public chatbot.
Example Prompts for the Data Agent
OpenAI provides several examples of questions users can ask.
A company could start with:
"Diagnose why weekly active users changed last week. Identify likely drivers, compare against previous periods and recommend the next checks."
For planning:
"Design a KPI framework for this new product area with primary metrics, drivers, guardrails, targets and data-validation requirements."
For leadership:
"Turn this month's metrics into a leadership-ready update with actuals, comparisons, drivers, caveats and recommended actions."
These examples show the intended direction of the product.
The goal is not simply to ask AI for a chart.
It is to give the agent an analytical problem and let it work through the underlying data.
Why OpenAI Data Agent Matters
The Data agent could become one of the more important enterprise applications of conversational AI because it addresses a common problem: companies have enormous amounts of data, but not everyone who needs that information knows how to work with it.
Traditional analytics tools solve part of this problem.
AI adds another layer by allowing employees to interact with business information through natural language.
The most important feature may therefore not be the dashboard generator.
It is the ability to ask a business question and investigate it interactively.
If the underlying data is reliable and the permissions are correctly configured, this could reduce the distance between a business question and a data-backed answer.
The Bigger Shift Toward AI-Powered Business Intelligence
OpenAI's Data agent also reflects a larger transformation in enterprise software.
AI is increasingly becoming an interface for existing business systems.
Instead of opening five applications and manually moving information between them, employees can increasingly ask AI to investigate information across connected systems.
That does not eliminate the underlying applications.
It changes how people interact with them.
For TheInfoBytes readers interested in AI tools, this is particularly important because it represents a shift from AI that creates content toward AI that works with company information and supports decisions.
OpenAI's new Data agent is an early example of that transition.
As enterprise AI develops further, similar systems are likely to become increasingly capable of connecting data, reasoning about business performance, creating reports and eventually taking approved actions based on the findings.
Frequently Asked Questions
What is OpenAI Data Agent?
OpenAI Data agent is an AI-powered analytics tool in ChatGPT Work that connects to approved company data, investigates business questions and creates interactive dashboards using natural-language instructions.
What data sources does OpenAI Data Agent support?
OpenAI lists sources including Amazon Redshift, Datadog, Google BigQuery, ClickHouse, Databricks, MongoDB, Snowflake, Google Drive and SharePoint, with additional integrations available through its ecosystem.
