Serverpod 4 - Tech preview

Today, we’re thrilled to share a tech preview of Serverpod 4. The first agentic coding engine that hot-reloads the full stack and can run locally on your computer. If you are a developer, this will forever change the way you think about building apps. We are finally closing the loop between your app's output, the backend, and your AI agent.

👉 Join us on Discord to be among the first ones to try it out.

Here is what makes Serverpod 4 radically different: Sub-second stateful hot reload across the server, database, web, and app. Everything is wired up to your favorite AI agent through a tried and tested MCP server and skills. We’ve added an embedded Postgres database; there's no need for complex setup or Docker containers. Everything just works.

You can literally see your app evolving as the agent works. No recompilation. No restart. No wait.

Serverpod 4 runs locally on your computer, and with Flutter for the frontend, your app runs at native speed on any platform. And did we say that it’s all free and open source?

What's in the package?

Servepod is an open-source, scalable backend written in Dart for the Flutter community. It provides end-to-end type safety all the way from the database to the client app, catching many errors before the app is even started. You can add a method on the server and call it from your app as if it were a local method.

New in Serverpod 4 is a complete full-stack runtime for agentic coding. One of Flutter’s core strengths has always been its stateful hot reload. We are now bringing this to the server and to the database. The serverpod start command will manage your full stack, including the Flutter app. Whenever you make a change to your code, database schema, or frontend, Serverpod will pick up the changes and instantly hot-reload them.

Your AI agent can interact with the running server through a new MCP server. It can create and apply database migrations, read logs, and even drive the app, taking and reviewing screenshots. The bundled rules and skills turn the AI agent into a senior Serverpod developer, and even lower-tier models understand how to write reliable, testable code.

We are still working hard to iron out the last couple of issues, but we are getting very close to a final release. We cannot wait to see what you will build with Serverpod.