Found a problem or have an idea? Write to the developer directly — replies usually arrive within 2 business days.
Send an emailpoy.utku2004@gmail.com
On purpose. AI Satranç ships with no chess engine: the AI's neural network is born with random numbers on first launch and only learns by playing. Its clumsy early games are proof that learning really starts from zero.
Open the Training tab and tap Start Training. The AI plays against itself and learns after every game — on a Mac this is the fastest way. It also learns from every game you play against it.
The Progress tab shows learning curves, training steps and match-test results. At regular intervals the new brain plays a tournament against its older self; if it has clearly gotten worse, the learning is rolled back.
On-device training uses the Metal GPU. Where Metal isn't supported (for example the iOS Simulator), the app can be played but does not learn.
Every player’s brain is connected to the shared brain: only a numeric summary of learning (a weight difference) and a random install number are shared, and what other players have learned reaches you too. Your games, moves and personal data never leave your device. In version 1.0 it can be turned off with Join the shared brain at the bottom of the Progress tab; in later versions it is on for everyone. If you are not signed in to iCloud, no contribution is sent. See the privacy policy for details.
Deleting and reinstalling the app fully resets the brain, archive and statistics — the AI is born random again.
No. No account, sign-in, ads, analytics or in-app purchases. Everything works offline.