HowAI Agent 🤖

A powerful AI chat app for iOS, Android, and Web powered by GPT-5.2 with built-in image generation, web search, voice synthesis, and more.

Features

Core Chat

• GPT-5.2 - OpenAI's flagship model with deep research mode
• GPT-5 Nano - Fast, lightweight model for free-tier users
• Conversation history - Local SQLite storage with cloud sync
• Multiple profiles - Separate chat histories per profile
• AI personalities - Customizable AI behavior and tone

AI Capabilities

• Image analysis - Analyze photos, screenshots, and documents
• Image generation - Built-in to GPT-5.2 (no separate model needed)
• Web search - Real-time internet search integrated into responses
• Deep research mode - High reasoning effort with comprehensive analysis
• Voice synthesis - Text-to-speech via ElevenLabs
• Document analysis - Process PDFs and files
• Presentation generation - Create PowerPoint slides from chat
• Places Explorer - Location-aware search with Google Maps & Street View

Premium Features

• Subscription tiers - Free and Premium plans
• Usage tracking - Weekly limits for free tier
• Cloud sync - Supabase-powered sync across devices

Personalization

• 17 languages - Full localization support
• Dark/Light mode - System or manual theme
• Font scaling - Adjustable text size
• Voice settings - Multiple ElevenLabs voices

Tech Stack

• Flutter - Cross-platform UI framework
• Supabase - Auth, database, and sync
• OpenAI - GPT-5.2 (chat, vision, image generation, web search)
• ElevenLabs - Text-to-speech
• Google Maps - Places API and Street View
• SQLite - Local data persistence

Project Structure

lib/
├── generated/       # L10n generated files
├── l10n/           # Localization ARB files
├── models/         # Data models (Profile, Conversation, etc.)
├── providers/      # State management (Provider)
├── screens/        # UI screens
├── services/       # Business logic and APIs
├── utils/          # Helper functions
└── widgets/        # Reusable UI components

haogpt-web/         # Web deployment (Docker)

Getting Started

Prerequisites

• Flutter SDK 3.0+
• Dart SDK
• API keys for: OpenAI, ElevenLabs, Google Maps, Supabase

Setup

1. Clone the repository:

   git clone https://github.com/hao6yu/howai-agent.git
   cd howai-agent

2. Copy env.example to .env and fill in API keys:

   OPENAI_API_KEY=sk-...
   # Optional in production: route through your backend proxy
   OPENAI_PROXY_BASE_URL=https://your-proxy.example.com
   OPENAI_PROXY_TOKEN=...
   ELEVENLABS_API_KEY=...
   GOOGLE_MAPS_API_KEY=...
   SUPABASE_URL=https://...
   SUPABASE_ANON_KEY=...
   

3. Install dependencies:

   flutter pub get

4. Generate localizations:

   flutter gen-l10n

5. Run the app:

   flutter run

Optional: OpenAI Proxy (Recommended for Production)

Use the included Supabase Edge Function proxy to keep OpenAI keys off-device:

1. Deploy function:

   supabase functions deploy openai-proxy

2. Set function secrets:

   supabase secrets set OPENAI_API_KEY=sk-... PROXY_SHARED_TOKEN=your-shared-token

3. Configure mobile app .env:

   OPENAI_PROXY_BASE_URL=https://<project-ref>.supabase.co/functions/v1/openai-proxy
   OPENAI_PROXY_TOKEN=your-shared-token

Building

# iOS
flutter build ios --release

# Android
flutter build apk --release

# Web (see haogpt-web/README.md)
cd haogpt-web && docker build -t haogpt-web .

Screens

Screen	Description
ai_chat_screen	Main chat interface
settings_screen	App configuration
profile_screen	Manage user profiles
ai_personality_screen	Customize AI behavior
subscription_screen	Premium plans
voice_settings_screen	ElevenLabs voice config
street_view_screen	Google Street View

Configuration

Key settings in SettingsProvider:

• useStreaming - Stream responses (disable for image gen)
• fontSize - Text scale factor
• selectedVoice - ElevenLabs voice ID
• enableHaptics - Vibration feedback

License

Private - All rights reserved

Support

☕ Buy me a coffee