Netlify Database Extends Deploy Preview Model to Postgres With AI-Native Branching.

Netlify has shipped Netlify Database, extending its decade-old deploy preview model to include the database layer. The platform now provisions a Postgres branch for every deploy preview, allowing developers and agents to test schema changes and data operations in complete isolation before touching production.

Distinguished Engineer Eduardo Bouças detailed the technical approach in a blog post published May 14, explaining how the team designed database branching as what they call an AI-native primitive. The launch follows Netlify's broader platform safety approach for agent-assisted development, which the company announced earlier this year.

Netlify introduced deploy previews in 2016 to give every feature branch its own isolated environment. Netlify Database extends that model by creating a database branch for each deploy preview, synced automatically with the application code branch. When a developer opens a pull request, the platform provisions both a preview environment and a corresponding database branch without additional configuration.

Migrations run automatically at build time, keeping schema and application code aligned across branches. The system applies migrations to preview branches first, then promotes them to production only after the pull request merges. This eliminates the manual coordination typically required when database changes span multiple environments.

Bouças notes the branching model solves a problem teams have worked around for years. "If you had 20 features in flight, you got 20 environments," he wrote, referring to deploy previews. "That model has covered almost every part of an application for a decade now, with one stubborn exception: the database, which stayed a single shared instance."

Netlify Database enforces strict access boundaries for AI agents. Agents running through Netlify's Agent Runners platform receive full read-write access to their assigned branch but cannot access production data. Local agents like Claude Code or Cursor receive read-only credentials by default, preventing accidental writes during local experimentation.

For local development, Netlify runs a WebAssembly build of Postgres that matches the production engine. Developers working on their laptops interact with the same database system that runs in production, reducing environment-specific issues. The Wasm approach also simplifies setup, since the database runs entirely in the local development environment without external dependencies.

Netlify Database is now generally available as part of the platform. The service provisions managed Postgres instances with automatic backups, scaling, and monitoring included. Teams can create database-backed projects using starter templates or add database support to existing sites through the Netlify dashboard.

The database service integrates with Netlify's existing build pipeline and framework support, including Next.js, Astro, and Remix. Pricing follows Netlify's existing tiered model, with database usage included in platform plans. The company has not disclosed technical specifications such as connection limits, storage caps, or geographic distribution of database instances.

Bouças positions the launch as infrastructure for what he describes as AI-native development, where agents participate directly in the development workflow rather than acting as external tools. "It's a good read on what AI-native actually means when operational safety matters as much as developer velocity," he wrote in the LinkedIn announcement.