M.A.D.E (Multi-Agent Decision Engine)

Inspiration

Students today aren’t lacking information, we’re overwhelmed by it.

From choosing a major to deciding on careers or graduate school, we’re constantly exposed to advice, data, and expectations. Instead of creating clarity, this overload often leads to confusion, overthinking, and inaction.

I realized the real problem isn’t access to information, it’s the inability to process it in a way that leads to confident decisions.

M.A.D.E. was inspired by a simple idea:
What if you could externalize your internal decision-making process and actually see your tradeoffs clearly?

What it does

M.A.D.E. (Multi-Agent Decision Engine) is an AI-powered decision platform that helps students make academic and career choices with clarity.

It simulates three internal perspectives:

Logical — focuses on practicality, risk, and long-term outcomes
Emotional — focuses on well-being, stress, and alignment
Ambitious — focuses on growth, opportunity, and future success

Instead of giving one answer, M.A.D.E. shows how these perspectives interact—helping users understand their internal conflict.

It then transforms that insight into a clear, actionable roadmap, including:

career recommendations
academic direction
salary projections
skill gaps
step-by-step next actions

How I built it

I built M.A.D.E. as a full-stack prototype:

Frontend: Next.js for a clean, responsive interface
Backend: Node.js to handle logic and data flow
AI System: A structured multi agent prompting system that simulates three distinct decision perspectives

User input is processed through each “agent,” and their outputs are combined into a structured response that feeds into a decision dashboard.

Challenges I ran into

Learning everything from scratch:
This was the first time my teammate and I worked with a full coding project like this. We were complete beginners with tools like GitHub, the terminal, and running full stack applications, so even getting started was a challenge.
Debugging and environment setup:
We ran into constant errors while trying to install dependencies, run the project, and understand how the frontend and backend connect. Learning how to navigate the terminal and troubleshoot issues took a lot of trial and error.
Connecting frontend, backend, and AI logic:
Getting all parts of the system to work together was one of the hardest parts. We had to figure out how data flows from user input -> backend -> AI logic -> back to the frontend.
AI integration limitations:
Due to limited API access, we couldn’t fully implement live AI responses. We worked around this by using mock data to simulate the intended experience.
Balancing learning with building:
At the same time that we were designing the product, we were also learning how to code, structure a project, and think like developers, which made the process both challenging and rewarding.

Accomplishments that I'm proud of

Building a working full stack prototype despite starting with little to no experience in tools like GitHub, the terminal, or full application development
Successfully combining AI, decision psychology, and career planning into one cohesive system
Designing a unique “Neural Twin” model that reflects real internal decision conflict
Turning a complex idea into a clear, user-facing product that provides actionable guidance, not just answers
Pushing through technical challenges and errors to bring the concept to life within a limited timeframe

What I learned

More information doesn’t solve decision making, structure and clarity do
Users don’t just want answers; they want direction and next steps
Building real products requires both technical problem solving and understanding human behavior
Debugging, trial and error, and persistence are a core part of development
Even as a beginner, it’s possible to build something impactful by learning quickly and staying consistent