Databricks API: what you can do and how to use it with AI agents
The Databricks REST API lets you manage jobs, clusters, SQL warehouses, notebooks and the data catalog, and run SQL queries. Data engineering, analytics and ML teams at mid-sized and large companies use it. Access is with a token or OAuth in the customer's workspace.
What the Databricks API offers
Jobs
Create, run and monitor pipelines and scheduled tasks.
Clusters and compute
Manage clusters and compute policies.
SQL warehouses and SQL execution
Run SQL queries and read the results.
Unity Catalog
Catalogs, schemas, tables and data permissions.
Workspace and notebooks
Manage notebooks, folders and repos.
Model serving
Published model endpoints for inference.
How to access the Databricks API
- With your company’s credentialsUses a personal access token or OAuth with a service principal, with permissions defined in the workspace and in Unity Catalog.
- Queries consume compute: set a warehouse and a cost ceiling for the agent.
- Use Unity Catalog permissions to limit sensitive data.
- Prefer a service principal to a personal token.
Official documentation: docs.databricks.com/api
What an iTeam agent does with the Databricks API
Business question in governed SQL
The agent turns a manager's question into a query on approved tables, runs it on the SQL warehouse and answers with the number and the query used.
Pipeline watch
The agent follows the overnight jobs and, if one fails, gathers the error and the history and alerts the responsible engineer.
Compute cost control
The agent flags idle clusters and oversized warehouses so the team can decide what to shut down.
Table access request
The agent takes the request, checks the policy and prepares the grant in Unity Catalog for the data owner to approve.
From the credential to the agent at work, with approval
Create the service principal
In Databricks, create a service principal with access only to the catalogs and warehouses it needs.
Store it in the iTeam vault
Add the credential and the workspace URL to the vault, keeping query agents separate from operations agents.
Validate queries and costs
Test on a small warehouse, review the generated queries and keep operational actions under approval.
Ready-made AI teams for this work
Management reporting team
Revenue, margin, pipeline and cash in one report, built from your systems, with a short note on what changed and why.
Data analysis team
Ask which customers stopped buying or which region lost margin. The team writes the SQL, runs it and explains the result.
IT service desk team
Password resets, access requests and how-to questions handled in chat, with Jira updated and approvals in place.
Questions about the Databricks API
Is the Databricks API free?
The API has no separate charge, but what it runs (clusters, warehouses, jobs) consumes DBUs and is billed according to your contract.
Do I need to write code to use the Databricks API with iTeam?
Not necessarily. The deployment team configures access and defines which tables and actions the agent uses. Your data team validates.
Can the agent change tables?
It can, if it is given permission, but we recommend starting read-only and leaving writes for approved cases.
