███████╗██╗  ██╗███╗   ███╗     █████╗ ██╗
  ██╔════╝██║ ██╔╝████╗ ████║    ██╔══██╗██║
  █████╗  █████╔╝ ██╔████╔██║    ███████║██║
  ██╔══╝  ██╔═██╗ ██║╚██╔╝██║    ██╔══██║██║
  ███████╗██║  ██╗██║ ╚═╝ ██║    ██║  ██║██║
  ╚══════╝╚═╝  ╚═╝╚═╝     ╚═╝    ╚═╝  ╚═╝╚═╝

A powerful, privacy-first, easy-to-access AI chat app for Android
Built with Jetpack Compose · Multi-model · No ads · No tracking

---

✦ Overview

ekm AI is a native Android AI chat application that connects directly to leading AI providers — Google Gemini and OpenAI-compatible APIs — with a beautiful, expressive Material You interface. No middlemen, no subscriptions, no servers. Your data stays on your device.

---

✦ Features

🤖 AI & Models

• Multi-model support — Switch between Google Gemini, OpenAI, and any OpenAI-compatible API
• Instant response — Get instant response for your queries
• Show Reasoning mode — Optionally appends the model's reasoning process below every response
• Global system prompt — Customize AI behavior across all conversations with a full-featured prompt editor
• Per-model accent colors — The entire UI tints to match whichever model is active

💬 Chat

• Markdown rendering — Full in-house markdown parser supporting bold, italic, strikethrough, inline code, fenced code blocks with language labels, headings, bullet & numbered lists, blockquotes, horizontal rules, and links
• Scrollable code blocks — Code never wraps or gets clipped; horizontal scroll with softWrap = false
• Copy button — One-tap copy on every AI message with animated ✓ feedback
• Model name label — Each AI bubble shows which model generated it, snapshotted at send time (no stale labels on scroll)
• Animated typing indicator — Three-dot wave animation while waiting for response

🎤 Voice Input

• On-device speech recognition — Uses Android's built-in SpeechRecognizer, no API key required, no usage limits
• Live RMS visualizer — Animated pulse rings driven by real microphone amplitude in real time
• Full voice flow — Listening → Processing → Transcript → Send, all in a beautiful bottom sheet
• Partial results — Transcript updates word-by-word as you speak

🔐 Security

• Key encryption — API keys encrypted at rest using c++ algorithm, never stored in plain text
• Zero telemetry — No analytics, no crash reporting services, no tracking of any kind
• No backend — All requests go directly from your device to the AI provider

🎨 UI & Theming

• Material You design — Expressive Material 3 components throughout
• 3 themes — Light, Dark, and System (follows device)
• Theme persistence — Selection saved to SharedPreferences, restored on every launch via Application class
• Full app restart on theme change — Clean process kill + relaunch via Process Pheonix lib for instant theme application
• Smooth transitions — Custom slide-up / slide-down / fade animations between all activities
• Edge-to-edge — Full edge-to-edge rendering with proper inset handling

⚙️ Settings

• System Prompt Editor — Monospace editor with live word/char counter, 6 preset templates (Default, Coder, Teacher, Writer, Concise, Friend), unsaved-changes indicator, and animated Save/Discard bar
• Show Reasoning toggle — Appends structured reasoning section to AI responses
• Theme picker — Full-page theme selector with visual chat previews per theme
• About page — App info, feature list, stats, team card
• Privacy Policy — In-app privacy policy with structured sections
• Cache clearing — One-tap cache clear with redirect to system settings for deeper cleaning

---

✦ Tech Stack

Core

Networking & Data

Architecture

Security

AI Providers

---

---

✦ Getting Started

Prerequisites

• Android Studio Hedgehog or newer
• Android device / emulator API 26+
• A Google Gemini API key & OpenAI-compatible API key

Setup

# 1. Clone the repo
git clone https://github.com/ekm-labs/ekm_android.git
cd ekm_android

# 2. Open in Android Studio
# File → Open → select the cloned folder

# 3. Build & run
./gradlew assembleDebug

---

✦ Permissions

Permission	Reason
INTERNET	API requests to AI providers
RECORD_AUDIO	Voice input via on-device speech recognition
ACCESS_NETWORK_STATE	Network availability indicator in chat

---

✦ Privacy

ekm AI has no backend infrastructure. Specifically:

• 🚫 No user accounts or sign-in
• 🚫 No analytics or crash tracking
• 🚫 No data sent to ekm servers (there are none)
• 🚫 No ads, ever
• ✅ All chat history stored locally on-device only
• ✅ All AI requests go directly to the provider you configured

When you send a message, it travels: Your device → AI Provider API → back to your device. That's it.

---

✦ License

The project is under MIT Liscense

---

Made with ♥ in India

ekm AI · ekm Labs · 2026