Local-First Flutter: How MindLeau Keeps Your Journal on Your Device

MindLeau is a brain-dump journal and mood tracker for iOS and Android. People write things in it that they would not post anywhere. That shaped the architecture more than any feature did: the default had to be that your words stay on your phone.

Why local-first for a journaling app

• Privacy by default. If nothing leaves the device unless you ask, there is much less to protect and much less to explain.
• It works anywhere. Guided sessions and journaling work without Wi-Fi, which is exactly when some people want them.
• It feels instant. Saving an entry is a local write, not a network round trip.
• No forced sign-up. Someone can open the app and start writing in seconds.

The stack

• Flutter for one codebase across iOS and Android.
• Drift on SQLite as the on-device source of truth. Drift gives typed tables and queries in Dart, migrations, and reactive streams, so screens update when the data changes.
• Firebase Auth and Cloud Firestore for optional background sync once a user creates an account.
• provider with ChangeNotifier and ValueNotifier for state.

How the data flows

The rule is simple: the UI only ever reads from the local database.

1. The user writes an entry. It is saved to Drift immediately.
2. The screen is listening to a Drift query stream, so it updates on its own.
3. If the user has an account, a sync layer pushes local changes to Firestore in the background and pulls remote changes back into Drift.
4. If the user has no account, step 3 never happens and nothing is lost from the experience.

Keeping the UI unaware of the network is what makes the offline experience reliable. There is no second code path for when the connection drops.

Why not Riverpod or Bloc?

Both are good tools. MindLeau did not need them. Most state in the app is "what is in the database right now", and Drift streams already provide that. The remaining UI state is small enough that provider with ChangeNotifier and ValueNotifier stays readable. Adding a heavier state management layer would have cost more in structure than it saved. Choosing the simplest tool that fits is a decision I make on purpose, not by default.

The trade-offs

Local-first is not free. The honest costs:

• Sync is your problem. A cloud-first app gets consistency from the backend. A local-first app has to decide what happens when the same data changes in two places.
• Schema migrations live on every device. Every database change needs a tested migration, because you cannot fix user data on a server.
• No account means no backup. Users who never sign in keep their data on one device. The app needs to say that clearly and make syncing an easy, optional step.
• Deletion has to be real. MindLeau deletes data instantly from Settings, which means deleting locally and, for synced users, remotely.

When I would choose local-first again

Local-first is a strong default for journals, notes, health and habit trackers, field tools, and anything used on unreliable connections. It is a poor fit for products that are social or collaborative at their core, where the shared server state is the product.

You can read the full MindLeau case study, or if you are planning an app and weighing architecture choices like this one, see how I help with technical product consulting.