Snowflake App Runtime

Snowflake App Runtime lets you go from an idea to a live, deployed web application in minutes. No infrastructure to provision, no credentials to configure, no Docker expertise required. Your app runs inside Snowflake, right next to your data, with direct access to your tables, warehouses, and security model. Describe what you want to build in Cortex Code CLI or Cortex Code Desktop, Snowflake’s AI coding agents. The agent scaffolds the project, wires up your Snowflake data, and deploys to a live URL.

Snowflake App Runtime builds Node.js apps (with a focus on Next.js). Python support is planned. It’s available in Amazon Web Services (AWS), Microsoft Azure, and Google Cloud commercial regions. It isn’t available in government regions or on trial accounts.

Why build on Snowflake?

  • Your data is already here. Query Snowflake tables directly from your application code. No API layers, no ETL, no data egress.
  • Enterprise-ready by default. SSO, role-based access control, audit logging, and governance are inherited from your Snowflake account. You don’t configure them; they’re already on.
  • Agentic build and deploy. Describe the app in Cortex Code; the agent scaffolds the project, helps you test locally, and deploys when you’re ready. You get a live, authenticated URL without managing Dockerfiles, container registries, or CI/CD pipelines.
  • Secure by design. Your app runs inside Snowflake’s security perimeter. End users authenticate through Snowflake’s existing identity provider. Queries can run as the calling user for per-user access control, with no auth system to build. No secrets to manage.

Quick start with Cortex Code

The fastest way to build and deploy is with Cortex Code CLI or Cortex Code Desktop. Both bundle the snowflake-apps skill; you invoke it a little differently in each client.

Describe your app in chat with the $snowflake-apps prefix:

> $snowflake-apps Build me a warehouse monitor that shows all my
  warehouses with their current state and credit usage today

Cortex Code generates a Next.js project, wires up Snowflake data access, tests locally, and deploys with snow app deploy. Your app is live in minutes.

See Getting started with Snowflake App Runtime for environment setup, the Snowflake CLI-only path, and connecting apps to Snowflake data.

Quick start with the CLI

If you already have application code, initialize and deploy:

snow app setup
snow app deploy

snow app setup generates an app.yml manifest. snow app deploy runs the upload, build, and deploy phases and returns a live Application Service URL. See Getting started with Snowflake App Runtime for the full walkthrough.

How this relates to Streamlit

Streamlit in Snowflake and Snowflake App Runtime are both ways to build applications directly on your data. They’re designed for different use cases:

  • Streamlit: Python-first, opinionated, guardrailed. The right choice for dashboards, data exploration, and analyst self-serve tools. Fast path to a polished data app without writing frontend code.
  • Snowflake App Runtime: Full power of the web. The right choice when you need custom UI, multi-step workflows, rich interactions, or the broader JavaScript and TypeScript ecosystem (React, Next.js).

Choose based on what you’re building. Both run inside Snowflake’s security perimeter, both inherit your RBAC, and both give you direct access to your data without building API layers. Both are part of Snowflake Apps: deployed apps surface in the same catalog regardless of how they were built.

For more about scaling apps and suspending them, see Scale and suspend Snowflake App Runtime apps.

Where to go next

The sidebar lists the rest of the doc set, grouped by task.