RizzMCP (WE MADE 70 DOLLARS)

TLDR

WE MADE 70 DOLLARS (pre product, post revenue)
IT WORKS (pre product, post revenue)

Inspiration

We noticed a lot of software engineers struggle in the dating world. Seriously, we all had friends who could code anything but couldn't get a date to save their lives. This got us thinking - why not use AI to help these technically-skilled but socially-awkward folks find love? That's how RizzMCP was born. We wanted to create an AI wingman that understands both code and romance.

What it does

RizzMCP is an AI agent that helps software engineers find romantic partners. It analyzes your personality, interests, and communication style to suggest potential matches. Then it coaches you through conversations, giving real-time feedback on your messages and helping you avoid common pitfalls like talking about code too much or forgetting to ask questions. It even suggests date ideas based on shared interests and handles scheduling so you don't have to worry about the logistics.

How we built it

We built RizzMCP using a custom MCP server architecture that integrates with dating apps through their APIs. The backend runs on Python with specialized NLP models trained to understand both technical jargon and romantic communication patterns. We implemented a windsurf cascade architecture to handle multiple user sessions simultaneously and ensure real-time coaching. The frontend is a simple React Native app that sits alongside your existing dating apps and provides suggestions without being intrusive.

Challenges we ran into

Getting the MCP server running was a nightmare. We spent three days straight debugging deployment issues that kept crashing our instances. Testing was also super difficult because we needed real dating app conversations to train our models but that data is super personal. We ended up having to create test personas and simulate entire relationships which took forever. Also explaining to our professors that we were building a dating assistant and not just watching Tinder all day was awkward.

Accomplishments that we're proud of

We actually made $70 in revenue during our beta test which is insane for a hackathon project. Several users reported significant improvements in their match rates and conversation quality. One guy even got a second date for the first time in years. Our windsurf cascade architecture successfully scaled to handle hundreds of simultaneous users which exceeded our expectations. The feedback system actually improved over time as it learned from successful interactions.

What we learned

We learned that building empathetic AI is way harder than technical AI. Getting the tone right for dating advice was super challenging especially since different people need different approaches. We also got better at distributed systems through all the MCP server issues we had to solve. The biggest lesson was probably about product market fit - there are so many engineers willing to pay for help with dating that we barely had to market the product at all.

What's next for RizzMCP (post product)

We're planning to expand beyond just engineers to help anyone who struggles with dating app conversations. We want to add video date coaching with real-time feedback through earbuds. The big goal is scaling our windsurf cascade architecture to support millions of users without latency issues. We're also exploring partnerships with dating apps to integrate directly rather than through their APIs. Long term we hope to use the conversation data to better understand what makes relationships successful from the very first message.