🧠 Team 4.4 – AI-Powered Feedback-to-Code Engine

24-Hour Project for iNTUition Hackathon 2025

🚀 Overview

Our project aims to redefine the software debugging and improvement process using real-time user feedback, vectorized data analysis, and AI-powered code transformation. This proof-of-concept demonstrates how developers can receive tested, relevant, and improvable code suggestions based on user pain points—saving hours of manual debugging and user analytics.

🎯 Problem Statement

Developers often spend significant time:

• Sifting through user complaints or reviews,
• Identifying the root causes of bugs,
• Manually testing fixes,
• Iterating before arriving at a final, stable solution.

We wanted to change that—by letting AI do the heavy lifting.

🛠️ What We Built

This system simulates a full loop that:

1. Collects and stores user feedback in SQL format.
2. Cleans and analyzes feedback using Python (TF-IDF + clustering).

3. Feeds structured prompts to OpenAI GPT to:

   • Identify bugs,
   • Suggest code improvements,
   • Test the changes (simulated),
   • Provide suggested output.

4. Developer Review UI:

   • If the AI-suggested output meets expectations, the developer can approve the code change with one click.
   • Otherwise, the developer can prompt the AI again to refine the solution.

🧩 Key Technologies Used

🧠 How It Works

graph TD A[User Feedback (SQL)] --> B[Python Data Cleaning] B --> C[Vectorisation (TF-IDF)] C --> D[Clustering Common Issues] D --> E[Prompt to OpenAI GPT] E --> F[Code Bug Identified & Suggestion Generated] F --> G[Test Build and Output Simulation] G --> H[Display Suggested Output to Dev] H --> I[Update Source Code] H --> E

Built With

• api
• python