yeahhe365/Dual-AI-Chat

What it solves

Dual AI Chat addresses the limitations of single-model AI responses by implementing a "collaborative intelligence" system. It prevents narrow perspectives or logical gaps by having two distinct AI personas—one focused on logic and one on creativity—debate and collaborate to produce more accurate and comprehensive answers.

How it works

The application uses a dual-agent architecture where every user query triggers a collaboration between two agents:

• Cognito (Logic Engine): Handles fact-checking, logical reasoning, structured analysis, and the final synthesis of the answer.
• Muse (Creative Engine): Focuses on lateral thinking, challenging assumptions, filling blind spots, and providing innovative perspectives.

These agents can share a "Smart Collaborative Notepad" that acts as an external memory, allowing both AIs to read and modify content. The system supports native Google Gemini integration and is compatible with any OpenAI-style API (such as DeepSeek, Ollama, or LM Studio), allowing users to assign different models to each persona.

Who it’s for

It is designed for users who need high-quality, deeply reasoned outputs for complex tasks and those who want to experiment with multi-agent collaboration and "thinking models" (CoT) without complex setup.

Highlights

• Dialectical System: Uses internal debate between logic-driven and creativity-driven agents to refine results.
• Thinking Model Support: Native integration for Gemini 2.5/3.0 with adjustable token budgets and thinking levels.
• Collaborative Notepad: A shared workspace with a Diff view to track additions and deletions in text or code.
• Flexible Backend: Supports official Gemini SDK and OpenAI-compatible endpoints for local or third-party LLMs.
• Multimodal Capabilities: Supports image uploads for visual-based AI discussions.