OpenAI and Databricks bring governed enterprise data to AI tools

OpenAI and Databricks are expanding their partnership to help enterprises use ChatGPT, Codex and GPT models with governed business data. The companies describe three main adoption paths: employee productivity, controlled AI coding and custom agents for business processes.
- OpenAI and Databricks' multi-year partnership began in September 2025.
- The Genie plugin connects ChatGPT to governed Databricks data and business definitions.
- Unity Gateway provides logging, budgets and model controls for Codex.
- Teams can deploy agents using GPT models, the OpenAI Agents SDK and Databricks Apps.
AI tools connected to company context
OpenAI provides the models and tools, while Databricks supplies the enterprise data, business definitions and access controls. The companies say this combination is intended to help AI understand how a business works and act on information that employees already govern in Databricks.
A central component is Databricks Genie Ontology, which stores approved metrics, business definitions and relationships between data. That context can help ChatGPT distinguish what terms such as “revenue” or “active customer” mean within a particular company. According to Databricks, users only see data they are already permitted to access, and responses link back to their sources in Databricks.
The partnership began in September 2025 as a multi-year agreement. Each company is also a customer of the other. Databricks says OpenAI is one of its largest customers, and that every Databricks engineer is getting access to Codex.
Three ways enterprises can use the partnership
The first path is business analysis through ChatGPT and the Databricks Genie plugin. Employees can ask questions in natural language, such as how weekly active user adoption is tracking against a target by plan tier. ChatGPT uses Genie to interpret the company's definitions, locate relevant data, run a query in Databricks and return results in graphs and charts. Users can also ask for recommended next steps, share findings through Slack or email, and take approved actions through connected tools.
The second path is coding with OpenAI Codex and Databricks Unity Gateway. The gateway lets companies provide Codex to engineers while retaining visibility into usage, spending and requests. Codex signs in with a Databricks credential, the OpenAI key remains in the gateway, and each request is logged. Engineering leaders can set budgets, review spending by team and choose which GPT model handles particular tasks.
The third path is custom agents for business processes. OpenAI frontier models can run next to company data through Databricks Model Serving under Unity Catalog governance. Teams can build and evaluate agents in Agent Bricks, or use the OpenAI Agents SDK and deploy agents as Databricks Apps. Databricks says Model Serving allows teams to call GPT models from SQL, notebooks and agents without managing a separate API key.
What the integration means for enterprise teams
The joint architecture assigns different roles to the products. ChatGPT is positioned as a place for employees to ask questions and complete work, Codex supports engineering tasks, and Databricks Apps hosts custom agents and applications. OpenAI models provide the reasoning across these uses, while Databricks products provide data access, governance and operational controls.
Unity Gateway routes model requests, applies team budgets and maintains an audit log. The Lakehouse stores company data under Unity Catalog governance, while Lakebase, a managed Postgres database, stores memory and state for agents. Other integrations include a ChatGPT Data Analyst plugin, a one-click ChatGPT connection to Databricks, Databricks Agent Skills for Codex and a Codex Data Science plugin.
For teams already using ChatGPT or Codex, Databricks recommends starting with the Genie plugin or Unity Gateway. Teams building inside Databricks can start with OpenAI GPT models there. Databricks says the same Unity Catalog permissions and Unity Gateway cost controls apply across these options, giving enterprises a common governance layer as they expand AI use.
Databricks: OpenAI and Databricks: Enterprise AI on your governed data

