Can I build and host an app entirely within my data platform?

Summary

• Yes. With Databricks Apps you build and host data and AI applications directly on the platform, running as containerized services on serverless compute, with no separate infrastructure to provision or manage.
• Use the frameworks your team already knows: Python frameworks (Streamlit, Dash, Gradio, Flask) or Node.js frameworks (React, Express, Angular, Svelte).
• Apps sit next to your governed data. They connect to Databricks SQL warehouses, Model Serving endpoints, and Lakebase (managed Postgres) without you managing credentials or connection strings.
• Security is inherited, not rebuilt. Users sign in with OAuth and OIDC through workspace single sign-on, and every app enforces the caller's Unity Catalog permissions with audit logging.
• Serverless hosting scales the app down when it's idle, and you deploy from the workspace UI or as code with Declarative Automation Bundles.

Can I build and host an app entirely within my data platform?

Yes. Databricks Apps lets you build and host data and AI applications directly on the Databricks platform, running as containerized services on serverless compute, with no separate servers or infrastructure to provision. Because an app runs inside your workspace, it sits next to governed data and models instead of calling out to a distant environment, and it inherits the platform's identity and access controls. Common apps include interactive dashboards, retrieval-augmented-generation chat assistants, data entry forms, and internal operational tools. See Introducing Databricks Apps.

What are Databricks Apps?

Databricks Apps are web applications hosted on the platform as containerized services. You develop an app locally or in the workspace, deploy it with the Databricks CLI or the workspace UI, and each app gets its own URL to share. The platform handles hosting, scaling, and maintenance, so there's no cluster to size and no web server to run. See the Databricks Apps documentation.

What to look for in an in-platform app hosting model

If the goal is to build and run an app without leaving your data platform, these are the capabilities that matter:

Capability	Why it matters
Hosting model	Serverless hosting removes infrastructure management and can scale an app down when it's idle.
Framework support	Support for common Python and JavaScript frameworks lets developers use what they already know.
Data and AI access	Direct connections to warehouses, model endpoints, and a transactional database keep app logic close to the data.
Identity and authentication	Built-in sign-in and single sign-on mean there's no separate auth layer to build.
Governance	Enforcing the caller's catalog permissions keeps apps inside existing data access rules.
Deployment and lifecycle	Deploy from a UI or as code, with source control and continuous delivery.

How Databricks approaches it with Databricks Apps

• Frameworks you already use. Build with Python frameworks including Streamlit, Dash, Gradio, and Flask, or Node.js frameworks including React, Express, Angular, and Svelte.
• Serverless hosting. Apps run on automatically provisioned serverless compute. Databricks handles hosting, scaling, and maintenance, and an app scales down when it's idle. See Databricks Apps is generally available.
• Direct access to data and AI. An app connects to Databricks SQL warehouses to query lakehouse data, to Model Serving endpoints for predictions and generative AI, and to Lakebase, the managed Postgres database built on the Neon serverless Postgres engine that is now part of Databricks. When you add Lakebase as a resource, Databricks provisions a service principal, grants it a matching Postgres role, and injects connection details, so the app connects without credentials in code.
• Security inherited from the platform. Apps authenticate users with OAuth and OIDC through workspace single sign-on, so there's no separate login to build. Each app runs with a service principal and enforces the user's Unity Catalog permissions, and access is captured in audit logs.
• Deploy from the UI or as code. Deploy in a few clicks from the workspace, or manage apps as code with Declarative Automation Bundles and continuous delivery from source control. See the Databricks Apps documentation.

Use cases

Teams build internal tools that would otherwise need a separate web stack: interactive dashboards and self-service analytics on live lakehouse data, chat assistants grounded in enterprise documents, data entry and data quality forms that write back to governed tables, and operational consoles that combine analytics, AI predictions, and transactional data in one interface. See building production-ready data and AI apps.

FAQs

What can I build with Databricks Apps?

Interactive dashboards, retrieval-augmented-generation chat assistants, data entry forms, self-service analytics, and internal operational tools. An app can combine governed lakehouse data, AI models, and transactional data in one interface.

Which frameworks are supported?

Python frameworks including Streamlit, Dash, Gradio, and Flask, and Node.js frameworks including React, Express, Angular, and Svelte.

Do I have to manage servers or scaling?

No. Apps run on serverless compute that Databricks provisions, scales, and maintains, and an app scales down when it's idle. You deploy the code and share a URL.

How do apps stay secure and governed?

Users sign in through workspace single sign-on with OAuth and OIDC, each app runs with a service principal, and apps enforce the caller's Unity Catalog permissions with activity captured in audit logs.