Hakoniwa AI

Shin Tomura

About this app

Hakoniwa AI is an educational tool designed for experiencing and learning the fundamentals of neural networks directly on your mobile device.

Concept
This app focuses on making the complex world of Artificial Intelligence accessible to everyone. By providing a visual and intuitive interface, you can understand how a neural network learns and makes predictions without any specialized knowledge or cloud-based processing.

Key Features

Full Offline Training
All calculations and learning processes are performed locally on your device. Since it does not require an internet connection or external servers, you can experiment with AI anytime and anywhere while ensuring complete privacy.

Intuitive Learning Process
You can input data, configure the network, and watch the training process in real-time. It is designed to help students and developers visualize how weights and biases change during the learning phase.

Pure Dart Engine
The app is built with an original engine written entirely in the Dart programming language. It does not rely on heavy external libraries like TensorFlow Lite, making it a lightweight and efficient tool for mobile environments.

Source Code Export for Developers
For those who want to take their experiments further, the app can export the trained model as a standalone Dart source code file. This code can be integrated into other Flutter projects, allowing you to use your trained models in practical applications without extra dependencies.

Who is this app for
This app is ideal for students interested in computer science, developers who want to understand the inner workings of AI, and anyone curious about how neural networks actually function.

Our Mission
Hakoniwa AI is a free tool with no advertisements. It was created to support the learning community and to provide a simple, transparent way to explore the possibilities of machine learning on mobile devices.

What’s new

Note: This app is a 100% offline AI learning simulator that runs completely on-device without using any external APIs.
・Added four math functions (log, exp, sqrt, abs) with fail-safe mechanisms to the formula builder and extended support to CSV batch prediction.
・Added dynamic histograms and category distribution bars to the data profiling UI.