README.md

metadata

title: MudabbirAI
emoji: 🧠
colorFrom: blue
colorTo: indigo
sdk: gradio
app_file: app.py
pinned: false
license: mit
tags:
  - agents
  - business
  - strategic-planning
  - mcp-1st-birthday

#mcp-in-action-track-2

#Link to the post: https://www.linkedin.com/posts/yhariri_mudabbirai-now-you-have-a-team-of-llms-working-activity-7400683779318566912-9quK?utm_source=share&utm_medium=member_desktop&rcm=ACoAAE5T1pcBL1zwsc5_5UW4QlINlrMaEfbTNSA

Team:

1. olihage_80597
2. youssefleb

MudabbirAI: The Strategic Selector

MudabbirAI is an advanced Multi-Agent System designed to solve complex, "wicked" business problems.

Instead of relying on a single LLM, it introduces Cognitive Diversity by assigning specific roles (Creative Plant, Pragmatic Implementer, Critical Monitor) to different AI models. It then runs a live Calibration Phase ("The Audition") to dynamically select the best-performing model (Gemini, Claude, Llama, etc.) for each specific role before generating a solution.

Key Features

• 🎭 Dynamic Team Recruitment: Auditions models in real-time to build the perfect team.
• ⚖️ The "Judge" & Self-Correction: Scores solutions on Novelty & Feasibility and forces self-correction loops if quality is low.
• 💰 Financial Intelligence: Tracks real-time token usage and cost for every step.
• 📊 Visual Analytics: Provides Radar Charts and Calibration Tables to visualize the decision-making process.

Used Models (Required more than 100k context minimum for the prompts)

• Google: Gemini 2.0 Flash
• Anthropic: Claude 3.5 Haiku 202421022
• SambaNova: Llama 4 Maverick 17B 128E Instruct
• OpenAI: GPT-4o-mini
• Nebius: Qwen 3 235B A22B Thinking 2507

How to Use

1. Enter your complex business problem in the text box.
2. Provide your API keys for the models you want to include in the audition (Google key is required for the Judge).
3. Click "Deploy MudabbirAI" and watch the system think, audition, and solve.

Research Papers that inspired the project

1. LLM-Creativity: Measuring and Explaining Creativity in Large Language Models: A Mixed-Methods Investigation into the Effect of Cultural Persona Prompts on Business Solution Generation (Youssef Hariri - 2025) https://doi.org/10.5281/zenodo.17494975
2. The LLM Team Composition Paradox: Why the Right Team Isn't Always Diverse: Developing a Contingency Model on the ARC-AGI Benchmark with an OODA-Belbin Framework Inspired by Multicultural Human Teams (Youssef Hariri - 2025) https://doi.org/10.5281/zenodo.17492778