What the Duck?!

AI-Powered Sprint Risk Detection, Anti-Pattern Analysis, and Rollover Prediction

What the Duck?! is an AI-driven SDLC infrastructure tool that analyzes sprint data, detects engineering anti-patterns, and predicts delivery reliability using probabilistic Monte Carlo simulation.
The goal is to make sprint planning predictable, reduce rollover percentage, and identify risks early before they become failures.

Features

AI Anti-Pattern Detection

Powered by Gemini 2.5 Pro. The system automatically identifies:

• Vague or incomplete task descriptions
  
• Missing acceptance criteria
  
• Tasks too large for a sprint
  
• Uneven workload distribution
  
• Dependency bottlenecks
  
• Scope-risk indicators
  

1,000-Run Monte Carlo Simulation

Predicts sprint outcome quality on a 0–100 scale:

• Worst case score
  
• Best case score
  
• Most likely score
  
• Full scoring distribution array
  

Noise factoring, clarity penalties, complexity variance, and rollover probability (commonly ~30%) are all modeled.

Zero-Upload Mode

The user uploads sprint data once.
All prediction pages automatically pull the most recent JSON from the database.
No repeated uploads required.
Press Predict and the system analyzes immediately.

Intelligent Sprint Insights

The system provides:

• Rollover likelihood
  
• Complexity distribution
  
• Bottleneck risks
  
• Probability of incomplete sprint delivery
  
• Clarity score and missing-info detection
  
• AI-generated improvement recommendations
  

Tech Stack

Frontend: Next.js, React, TailwindCSS, Framer Motion
Backend: Next.js Route Handlers, Prisma, MongoDB
AI: Gemini 2.5 Pro (Google Generative AI)

Built With

• framer-motion
• gemini
• mongodb
• nextjs
• prisma
• react
• tailwind
• typescript