Evgenii Novikov

Shopify went all-in on React Native in 2020, finished moving its apps last year — and in September 2026 announced it is going back to native Swift and Kotlin. This is not our stack, but the reason is coding agents, and agents change what matters when you pick a framework at all. I build a fullstack Dart framework, and I already know I will not be the one writing apps on it: agents will. That changes the price of every technical choice, and Shopify is a clear case of it.

An MCP server has been built into the Dart SDK since July 2025, and only recently have people started actually using it. MCP is the protocol that lets a coding agent call a server for information or actions. For Dart that means the agent is no longer limited to the terminal — it can reach into your running app.

We connected it to Claude Code on a demo app and checked what it gives in practice.

Serverpod 4 came out in September, and with it App Studio. We have moved our own framework off Serverpod, but for fullstack Dart it remains a solid choice — so here is the release, feature by feature, with what each one is and what it is worth in practice.

The lens matters, so it goes first. Most of our development no longer happens in front of a running app. One session holds the conversation about the work; it starts separate agent sessions for development and for code review; several tasks run at once, in the background; and the result gets looked at on a test server a couple of times a day. Several of Serverpod 4's headline features are built for a different picture — a developer at a desk with the whole stack running locally.

Summer produced exactly one stable Flutter release. 3.45 and 3.46 never reached stable at all, and 3.47 arrived on 12 August carrying 1358 commits on top of 3.44. The headline is not a feature. It is a repackaging: material and cupertino moved out of the SDK into separate packages.

That sounds alarming and mostly is not — today. Here is what we found migrating a real project from 3.44.0 to 3.47.2, and the one widely repeated claim about this release that does not survive contact with the code.

Letting people sign in with either a phone number or an email looks like a one-line change: widen the lookup, accept both. We shipped it last week in a live project, and the one-line change turned out to be a data-model change with a bill attached.

Here is what it actually costs, in the order the costs show up.