Pulse

A "good" artificial intelligence made by just one person.

Structure

• include/pulse, src — core neural network engine, written in pure C++17 with no external dependencies (Tensor, Layer, DenseLayer, Network, Trainer).
• platform/android — Android app module with a JNI bridge (PulseBridge) exposing the core engine to Java, plus an animated chat home screen (MainActivity, PulseWaveView, TypingIndicatorView, PulseResponder).
• platform/ios — iOS static library target with an Objective-C++ bridge (PulseBridge), plus a SwiftUI chat home screen under platform/ios/App (ChatView, PulseWaveView, TypingIndicatorView, PulseResponder).
• assets/icon — app icon source and exported PNGs.
• .github/workflows/build.yml — CI building Windows, Linux, macOS, Android and iOS.

Build (desktop)

cmake -S . -B build
cmake --build build --config Release

Build (Android)

cd platform/android
gradle assembleDebug

Build (iOS)

cmake -S platform/ios -B build-ios -G Xcode \
  -DCMAKE_TOOLCHAIN_FILE=cmake/ios.toolchain.cmake \
  -DPULSE_IOS_PLATFORM=SIMULATORARM64
cmake --build build-ios --config Release

This only builds the static library. Xcode project files (.xcodeproj) are binary/generated artifacts that Xcode itself must create, so to run the chat app on iOS:

1. In Xcode, create a new iOS App project (SwiftUI, bundle id com.pulse.app).
2. Add all files from platform/ios/App to the project.
3. Add PulseBridge.h, PulseBridge.mm and the pulse core sources/headers to the target, and set platform/ios/App/Pulse-Bridging-Header.h as the target's Objective-C bridging header.
4. Replace the generated Assets.xcassets with platform/ios/Assets.xcassets and the Info.plist with platform/ios/Info.plist.
5. Build and run.