ArgueMate

An AI agent that helps users practice and sharpen their IPDA debate skills in a fun and engaging way with AI-powered practice, topic guidance, and real-time feedback.

Comment

Inspiration

The inspiration for ArgueMate came from the difficulty of practicing speech and debate. Consistent practice requires multiple people, a judge, and a structured format like IPDA, which makes it difficult for individuals to practice on their own.

We also noticed that IPDA resources online are scattered and not beginner-friendly. Even when turning to AI tools, they often require constant prompting to follow the correct format, roles, and structure. This creates unnecessary friction and takes focus away from actually improving debate skills.

We wanted to build something that removes that friction and allows users to focus purely on debating. This need became even more clear through firsthand experience within our team, where one of our members participate in IPDA debate her first year at UCSD and struggled to find resources for getting enough structured practice.

What it does

ArgueMate is an AI-powered debate platform that simulates a full IPDA-style round. It allows users to practice anytime without needing another person or judge.

The AI automatically follows the debate structure, plays the role of opponent and judge, and progresses through each stage of the round. Users can focus on building arguments, responding to opponents, and improving their skills without worrying about formatting or setup.

The site as a whole also allows user's to look up school's in their city with Speech and Debate teams, and has a minigame, Fact CheckMate that helps them get practice identifying strong vs. weak pieces of evidence which is a crucial skill when debating.

How we built it

  • Next.js (React) – For a fast, server-side rendered frontend with smooth client-side navigation.
  • CSS – For styling and creating a clean, responsive interface.
  • JavaScript & TypeScript – For type-safe and flexible application logic.
  • Puppeteer – For automated web scraping to gather info about university speech and debate teams.
  • Gemini API – To generate structured debate arguments and strong/weak pieces of evidence to quiz the user with.

Challenges we ran into

One of the biggest challenges was getting the AI to consistently follow IPDA structure. It often required extensive prompt tuning to keep responses organized and on track.

We also faced difficulties integrating the API and ensuring smooth communication between the frontend and backend as it was both our first times using an AI API.

Another challenge was formatting responses clearly on the frontend and extracting useful, structured data from scraped sources.

Accomplishments that we're proud of

We’re proud that we were able to build a fully functional debate system that properly follows the IPDA structured format. Creating an experience where users can jump straight into a debate without setup was a major achievement.

We’re also proud of how much we improved the AI’s consistency through prompt engineering, turning it into a reliable debate partner and judge.

Finally, we’re proud that we were able to build a working project in just over a day from a simple idea as beginners, and even implement additional features we hadn’t initially expected to complete, such as Fact CheckMate and the school search.

What we learned

Building ArgueMate gave us hands-on experience working with an AI API and connecting the frontend and backend into a complete, functional application. We also learned prompt engineering, figuring out how to guide AI to follow specific rules and formats.

We improved our skills in building responsive interfaces, handling real-world data, and creating user-focused experiences that are clear and easy to use. As beginners, we also learned how to work effectively as a team: dividing tasks, sharing ideas, and managing a small project from start to finish.

What's next for ArgueMate

  • Immediate Grading: Evaluate user responses as they type, highlighting text boxes red for very weak points, yellow for weak points, and blue for strong points. Also provide small suggestions throughout the debate instead of waiting until the end.
  • Expanded debate formats: Support additional debate styles beyond IPDA, such as NPDA, Lincoln-Douglas, and Policy Debate.
  • Interactive learning games: Add more games like Fact Checkmate to make learning debates more engaging and fun.
  • Deployment!

Built With

Share this project:

Updates