🤖 Swift AI Agent Demo

🚀 A powerful iOS implementation of the ReAct (Reasoning + Acting) Agent pattern, bringing intelligent AI reasoning directly to your mobile device!

📱 Screenshots

✨ Features

🧠 Intelligent Reasoning: Watch the AI agent think through problems step-by-step 🔧 Dynamic Tool Usage: File operations, data processing, and more 📱 Native iOS UI: Beautiful, responsive interface optimized for mobile ⚡ Real-time Updates: Live visualization of the agent's thought process 🔄 ReAct Pattern: Implementation of the cutting-edge Reasoning + Acting paradigm 🎯 User-Friendly: Clear visual hierarchy showing each step of the reasoning process

🏗️ Architecture

This project implements the ReAct (Reasoning + Acting) pattern, where the AI agent:

1. 🤔 Thinks - Analyzes the problem and plans the next step
2. 🎯 Acts - Uses available tools to gather information or perform actions
3. 👀 Observes - Processes the results from the action
4. 🔄 Repeats - Continues until the task is complete

📦 Core Components

• 🌐 ChatGPTService: OpenAI API integration for intelligent responses
• 🤖 AgentService: Core ReAct logic and tool execution engine
• 🎨 ContentView: Beautiful UI displaying the reasoning process
• 📱 ContentViewModel: State management and UI coordination
• 🔧 Tool System: Extensible tool framework for agent capabilities

🚀 Quick Start

Prerequisites

• 📱 iOS 18.0+ / macOS 14.0+
• 🛠️ Xcode 16.0+
• 🔑 OpenAI API Key

Installation

1. Clone the repository

   git clone https://github.com/banghuazhao/swift-ai-agent-demo.git
   cd swift-ai-agent-demo

2. Open in Xcode

   open ReActAgent.xcodeproj

3. Configure API Key

   • Open ChatGPTService.swift
   • Replace the API key with your OpenAI API key

   private let apiKey = "your-api-key-here"

4. Build and Run 🏃‍♂️

   • Select your target device/simulator
   • Press Cmd + R to build and run

🎮 How to Use

1. 📝 Enter Your Query: Type any question or task in the input field
2. 🚀 Start the Agent: Tap "Start Agent" to begin the reasoning process
3. 👀 Watch the Magic: Observe as the agent thinks, acts, and learns
4. ✅ Get Results: Receive intelligent, step-by-step solutions

💡 Example Queries

🧠 Simple Tasks

• "What is the current time?"
• "Calculate 15 + 27"
• "Create a shopping list and save it to a file"

🤔 Multi-Step Reasoning Tasks

These examples demonstrate the ReAct pattern by requiring the agent to:

1. Think about the problem and break it into steps
2. Act by using tools (calculate, write_file, read_file, get_current_time)
3. Observe the results from each action
4. Repeat the cycle until the task is complete

• File Operations: "Create a file called 'notes.txt' with some important notes, then read it back to me"
• Data Processing: "Calculate the sum of 25, 37, and 48, then save the result to a file called 'calculation.txt'"
• Time-Based Tasks: "What time is it now? Then create a file called 'timestamp.txt' with the current time"
• Complex Calculations: "Calculate (15 * 3) + (27 / 9) and save the result to a file"
• Multi-File Operations: "Create a file called 'todo.txt' with a todo list, then create another file called 'summary.txt' that says 'Todo list created successfully'"
• Conditional Reasoning: "Get the current time, then create a file called 'time_log.txt' with the message 'Current time logged'"
• Data Analysis: "Calculate the average of 10, 20, 30, 40, and 50, then save both the individual numbers and the average to a file called 'statistics.txt'"
• Workflow Tasks: "Create a file called 'workflow.txt' with the text 'Step 1 completed', then read the file to confirm it was created"
• Mathematical Operations: "Calculate 100 divided by 4, then multiply the result by 3, and save the final result to 'math_result.txt'"
• Information Gathering: "What time is it? Then create a file called 'time_report.txt' with the current time and a message saying 'Time report generated'"

🧪 Advanced Testing Examples

For developers wanting to test the agent's reasoning capabilities:

• Multi-Step Math: "Calculate 50 * 2, then add 25 to the result, then divide by 5, and save the final answer to 'complex_math.txt'"
• File Chain: "Create file 'step1.txt' with 'Hello', then create 'step2.txt' with 'World', then read both files and create 'combined.txt' with their contents"
• Time Series: "Get the current time, wait a moment, then get the time again and create a file called 'time_comparison.txt' with both timestamps"
• Data Transformation: "Calculate the square of 12, then calculate the cube of 3, then add them together and save to 'transformed_data.txt'"
• Validation Workflow: "Create a file called 'test.txt' with 'Test content', then read it back to verify it was written correctly"

🛠️ Available Tools

Tool	Description	Usage
📄 read_file	Read contents from files	read_file("filename.txt")
✍️ write_to_file	Write content to files	write_to_file("output.txt", "content")
📅 get_current_time	Get current date and time	get_current_time

🔧 Technical Details

🏛️ Architecture Pattern

• MVVM: Clean separation of concerns
• Observable: Reactive UI updates
• Async/Await: Modern concurrency handling
• Service Layer: Modular, testable components

🎨 UI Components

• SwiftUI: Declarative, modern UI framework
• Custom Views: Tailored components for agent visualization
• Responsive Design: Optimized for all iOS devices
• Accessibility: Full VoiceOver support

🔐 Security & Privacy

• Local Processing: Agent reasoning happens on-device
• Secure API: Encrypted communication with OpenAI
• No Data Storage: Conversations are not persisted
• Privacy First: Your data stays private

🤝 Contributing

We welcome contributions! Here's how you can help:

1. 🍴 Fork the repository
2. 🌟 Create a feature branch (git checkout -b feature/amazing-feature)
3. 💾 Commit your changes (git commit -m 'Add amazing feature')
4. 🚀 Push to the branch (git push origin feature/amazing-feature)
5. 📥 Open a Pull Request

🐛 Bug Reports

Found a bug? Please create an issue with:

• 📱 Device/iOS version
• 🔍 Steps to reproduce
• 📸 Screenshots (if applicable)
• 📋 Expected vs actual behavior

📚 Learn More

🔗 Related Resources

• ReAct Paper - Original research paper
• OpenAI API Documentation - API reference
• SwiftUI Documentation - UI framework

🎓 Educational Value

This project serves as an excellent learning resource for:

• 🤖 AI Agent development
• 📱 iOS/SwiftUI development
• 🔄 Reactive programming patterns
• 🏗️ Clean architecture principles

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• 🧠 Inspired by the MarkTechStation
• 🔧 Built with OpenAI's powerful GPT models
• 🎨 Designed with Apple's Human Interface Guidelines
• ❤️ Made with passion for AI and mobile development

📞 Contact

Banghua Zhao

• 🐙 GitHub: @banghuazhao
• 📧 LinkedIn: https://www.linkedin.com/in/banghuazhao/

---

⭐ Star this repository if you found it helpful!

Building the future of mobile AI, one agent at a time 🚀