Dual AI Chat: Debate-Style AI Q&A App 180

Dual-ai-Chat is a chat application whose core design is to have two AI systems with different approaches (Cognito and Muse) collaborate through debate to generate answers. By default, it connects to the Google Gemini API, while also being compatible with other OpenAI-standard API interfaces, and supports local models or other hosted services.

Key Features:

• Dual-AI Debate: Cognito (logic-oriented) and Muse (skepticism-oriented) first discuss the question internally, reducing misinformation through cross-verification and analyzing from multiple perspectives before outputting the final answer.

• Multi-Interface Support: Natively compatible with Google Gemini, while also supporting all OpenAI-standard interfaces, allowing flexible switching between local models or third-party services.

• Shared Notepad: Both AIs share a Markdown-enabled notepad, with multi-step undo/redo for all discussions and edits.

• Multimodal Input: Supports uploading images and text; the AI can identify the image content and incorporate it into the discussion.

• Flexible Configuration: Freely switch AI models, adjust interface settings, and customize discussion modes and prompts.

• Error Handling: Automatically retries failed requests and provides a manual retry button to ensure uninterrupted conversation.

Technical Implementation:

• Frontend developed with React 19 + TypeScript

• Build tool is Vite

• Styling implemented via Tailwind CSS (loaded via CDN)

• Dependency management loads ES modules directly through Import Map in index.html, no local node_modules required

• Icon library uses Lucide React

• Markdown rendering handled by Marked, with security sanitization by DOMPurify

Development Environment Setup:

1. Requires Node.js 18 or higher

2. Clone the repository: git clone <repository URL> && cd dual-ai-chat

3. Install dependencies: npm install

4. Start development mode: npm run dev; the browser will automatically open the local access URL

How to Use:

• Enter a question: Supports text input or uploading an image to ask

• View the process: Observe the discussion details between Cognito and Muse through chat bubbles

• Get the answer: The final answer is displayed in the notepad

• Notepad operations: Supports preview/source toggle, full-screen display, content copy, and multi-step undo/redo

• Personalization: Adjust AI backend, discussion mode, interface font size, and system prompts through the settings panel

Use Cases:

• Learning & Research: Analyze academic questions and resolve learning difficulties

• Content Creation: Optimize copywriting and spark creative inspiration

• Professional Work: Troubleshoot technical issues and assist with business decisions

• Daily Assistant: Travel and renovation consultation, casual chat

Current Limitations:

• Replies require waiting for complete generation; word-by-word streaming output is not supported

• In "AI-driven" mode, the two AIs may repeatedly debate each other

• All AI requests are executed sequentially; you must wait for the current task to complete before initiating a new request