_____      _      ____    ____  
/  ___|    / \    |  _ \  / ___|
| |       / _ \   | |_) | \___ \ 
|  ___   / ___ \  |  _ <   ___)
 \____| /_/   \_\ |_| \_\ |____/

 Unleashing collective intelligence through structured multi-AI collaboration

CARS is a comprehensive framework for conducting deep, thorough research using multiple AI models working collaboratively with a human operator. It structures the research process to leverage diverse AI perspectives while reducing cognitive load on the human operator.

Core Principles:

• Human-in-the-Loop: The operator is the central decision-maker who provides context and verifies results
• AI Diversification: Utilizes multiple AI models with different strengths to avoid confirmation bias
• Iterative Validation: Employs a cyclical process with continuous verification and refinement
• Structured Approach: Follows clear steps with defined outcomes at each stage
• Quantitative Assessment: Uses metrics where possible and appropriate
• Contextualization: Quality depends on the completeness of operator-provided context

• Overcome Cognitive Limitations: Reduces blind spots through multi-perspective analysis
• Increase Research Depth: Promotes thorough exploration of the problem space
• Enhance Critical Thinking: Structured critique helps avoid confirmation bias
• Improve Synthesis Quality: Deliberate integration of diverse viewpoints leads to better conclusions
• Reduce Operator Dependency: Clear prompts and procedures make the process less dependent on operator expertise

The CARS framework follows a phase-based approach with clear checkpoints:

• Defining the research topic/problem
• Activating the AI consortium

• Generating clarifying questions from multiple perspectives
• Providing context and answers
• Formulating the final research task statement

• Identifying specific sources and queries
• Collecting relevant information
• Structuring and pre-processing collected information

• Conducting hypothesis brainstorming from multiple perspectives
• Critically evaluating and prioritizing hypotheses
• Selecting hypotheses for in-depth research

• Decomposing hypotheses and planning verification
• Targeted information collection
• Analysis and interpretation of data
• Cross-validation and consensus building
• Decision-making on hypothesis status

• Aggregating results across all hypotheses
• Connecting findings to research questions
• Documenting strengths, weaknesses, and limitations

The framework leverages four distinct AI perspectives:

• Analytical: Focuses on data, facts, logic, structure, and metrics
• Critical: Identifies weaknesses, risks, limitations, biases, and alternative explanations
• Creative: Generates ideas, hypotheses, non-standard approaches, and new possibilities
• Synthesizing: Integrates information, formulates conclusions, generalizes, and finds consensus

• Access to multiple AI models (recommended models as of 04.2025):
  • Gemini 2.5 pro
  • Claude 3.7 sonnet (thinking)
  • Grok3 thinking / research
  • Deep Seek r1
  • Open AI O3 + deep research

1. Clone this repository
2. Review the framework documentation
3. Set up your AI access points
4. Define your research topic
5. Follow the structured prompts in each phase

One of the key features of CARS is tracking explored vs. unexplored areas of the research space:

┌───────────────────────────────────────────────┐
│              RESEARCH SPACE MAP               │
├────────────────┬─────────────┬────────────────┤
│  EXPLORED      │ FRONTIER    │  UNEXPLORED    │
├────────────────┼─────────────┼────────────────┤
│ ▓▓▓▓▓▓▓▓▓▓▓▓▓▓ │ ▒▒▒▒▒▒▒▒▒▒▒ │ ░░░░░░░░░░░░░░░│
│ ▓▓▓▓▓▓▓▓▓▓▓▓▓▓ │ ▒▒▒▒▒▒▒▒▒▒▒ │ ░░░░░░░░░░░░░░░│
│ ▓▓▓▓▓▓▓▓▓▓▓▓▓▓ │ ▒▒▒▒▒▒▒▒▒▒▒ │ ░░░░░░░░░░░░░░░│
├────────────────┼─────────────┼────────────────┤
│ High           │ Medium      │ Low            │
│ Confidence     │ Confidence  │ Confidence     │
└────────────────┴─────────────┴────────────────┘

CARS is designed to be self-improving through meta-cognitive loops. Each application of the framework provides insights for its own enhancement.

We welcome contributions to improve the framework! Please see our contribution guidelines for more information.

• Special thanks to all the AI models that collaborate in this framework
• Inspired by methodologies in systems thinking, metacognition, and collaborative research

"The whole is greater than the sum of its parts." - Aristotle