Harry Cartwright - Portfolio

Case Study - Layla

Private AI chatbot that operates directly on your device without requiring an internet connection. Offline AI with LLMs and Stable Diffusion models.

Industry
Mobile App
Year
Service
Mobile App (iOS)

About

Layla is a groundbreaking personal AI that resides directly on your phone or device. No internet connection required, no censorship, complete privacy. No information leaves your device.

The app downloads a 4GB Large Language Model (LLM) after it starts for the first time, enabling fully offline AI capabilities.

Features

  • General Knowledge: Ask Layla about anything under the sun, she will give you an informed response

  • Multiple Personalities: Choose from many personalities - from cute to serious to bubbly to stoic

  • Memory: Tell Layla about yourself, she'll remember you

  • Personalisation: Create your own characters if none of the presets fit your style

  • Voices: Over 100+ different voices to choose from for your characters

  • Roleplay: Engage in roleplay with multiple characters

  • Task Scheduling: Basic task scheduling/reminder features - just talk to Layla normally

  • Custom AI: Load custom GGUF models that fit your needs

  • Free Plugins: Additional features available as free plugins including roleplay, downloading different personalities, and advanced settings

Solution

High level architecture of the application.

Layla is built using modern Android development practices with Kotlin and Jetpack Compose for the UI. The app implements MVVM architecture pattern and uses Room Database for local storage. Kotlin Coroutines handle asynchronous operations, ensuring smooth performance even when processing large language models locally.

The app supports loading custom GGUF models and integrates Stable Diffusion for image generation, all running completely offline on the device.

Visit website

Technologies

Kotlin

Java

Android Studio

Android SDK

Jetpack Compose

Material Design

Room Database

Coroutines

MVVM

More Applications

main*
Go Live