GameNet

• 

  example of leinad chat

• 

  what searching for a game looks like

• 

  sharing with others via shareable link

• 

  another recommendation example

• 

  example recommendation based on some games

Inspiration

Are you a gamer? Well, so are we! Video games are a great way to spend free time and foster a community, however, finding new games can be challenging. Recommendation platforms exist, mostly through large companies like steam, however, they don't actually give explanations for why we should play game x over game y, and even have biases in how games will be recommended, by recommending popular games over a users specific preferences. This then requires users to put in extra work in finding the games they want to play. This is why we decided to make GameNet, an agentic AI system that recommends players similar games to ones that they've selected as their favorites. GameNet is a tool for gamers to find the games they want to play, not what a corporation wants them to play.

What it does

GameNet takes information from the RAWG API, which contains a list of games, genre's, and other datatype features like screenshots. We used this information to set up an Agentic AI recommendation system(Gemini Based). That allows users to select up to 5 games(they can do less), and then get returned recommended games based on ratings(Metacritic) and genre.

Alongside that, users are also able to select the games and see screenshots of what the game looks like. Furthermore, we also have an interactive chatbot that's also based in Gemini. Where users are able to ask why these games are being recommended to the user. This allows users to have conversations to better understand whether a game is a good fit for them or not. Something, that even steams recommender system doesn't have.

How we built it

We built this project using JavaScript primarily. After that we focused on setting up our API Keys and setting up our Webpage.

Challenges we ran into

Setting up the RAWG API and integrating it into our project. After that we also had a challenge with Google Gemini's API keys but that was more of a UI issue on their end. Alongside that we also we're testing to make sure that our recommendations being formulated by Gemini's LLM we're actually accurate and we needed to change the prompting a few times.

What we Learned & Accomplished

We learned about the RAWG API, how to create an Agentic AI based system using Gemini. As well as increasing our skills when it comes to using LLM's to solve a practical problem.

What's next for GameNet

Fixing any bugs that pop up within the chatbot/recommender system. Alongside adding a feature that allows users to share their recommended games.

Built With

• gemini
• git
• github
• javascript
• nextjs
• react
• tailwind
• vercel