Future Compress

🚀 Inspiration

Modern decision-making is fundamentally flawed. People make life-changing choices—education, business, finance—without understanding long-term consequences.

Most tools today provide static data or a single prediction. None allow users to explore multiple possible futures before committing to a decision.

As a solo developer, I wanted to challenge this limitation by building a system that transforms uncertainty into clarity.

Instead of answering: "What should I do?"

I built something that answers: "What will likely happen if I do this?"

💡 What it does

Shadow Decisions is an AI-powered decision intelligence platform that simulates multiple possible future outcomes based on a user’s input.

Users can:

Input real-world decisions in natural language
Generate multiple future scenarios (best, worst, most likely)
View probabilities, risks, and impact metrics
Explore an interactive decision tree
See how outcomes evolve over time
Adjust parameters like risk tolerance and budget in real time
Compare different decisions side-by-side
Receive AI-generated explanations and recommendations
Analyze risk through visual heatmaps
Track a “regret score” for each decision

Instead of giving one answer, the system reveals multiple possible futures and helps users understand the consequences of their choices.

⚙️ How I built it

Backend

Python + FastAPI for API development
Modular architecture separating simulation, ML logic, and routing
Monte Carlo simulation engine to generate multiple future scenarios
Custom prediction logic for probability, risk, and impact scoring

Machine Learning

NLP-based scenario parsing (structured from user input)
Feature engineering for risk and decision variables
Probabilistic modeling to simulate uncertainty
Outcome scoring system for ranking futures

Frontend

Next.js (React) for dynamic UI
Tailwind CSS for styling
Framer Motion for animations
Interactive components for results, decision trees, and metrics

Visualization

Dynamic charts for probabilities and outcomes
Decision tree representation for multi-path futures
Risk heatmaps and timeline evolution views

⚠️ Challenges I ran into

Designing a system that produces multiple realistic outcomes instead of a single prediction
Simulating real-world uncertainty without access to full real-time datasets
Structuring decision trees in a way that is both interactive and understandable
Balancing technical complexity with user-friendly design
Managing full-stack development alone under time constraints

🏆 Accomplishments that I'm proud of

Building a complete AI-powered system as a solo developer
Creating a multi-outcome simulation engine instead of a basic predictor
Designing an interactive decision tree for exploring future scenarios
Delivering a visually engaging and dynamic user experience
Turning a complex concept (decision intelligence) into a usable product

📚 What I learned

How to design systems that simulate uncertainty instead of relying on deterministic outputs
Integrating machine learning logic into real-world applications
Building scalable full-stack architectures independently
Creating meaningful user experiences around complex data
Thinking beyond "features" and focusing on impact and usability