EcoQuest

• 

  Google ADK Agent Architecture

Inspiration

Inspired by AI's potential as an educational tool, particularly in raising awareness of its applications, we developed an educational game themed around marine creatures. The game promotes ecological awareness while addressing a common challenge faced by game developers: repetitive coding. This issue is effectively solved through our Google ADK agents powered by the Gemini API, reducing overhead and redundancy.

What it does

Our backend agents dynamically generate game content, handling the workload and maintenance that comes with game development. EcoQuest is the driver of this technology.

How we built it

We combined Gemini's API with Google's ADK to create agents with specific tasks. Some agents were tasked with interpreting the current game state and sending the tasks to other agents, whose task were then to generate JSON data depending on the instructions we prompted them.

Challenges we ran into

Bridging the front end with our backend proved to be very difficult, and having them work together required a large amount of effort.

Accomplishments that we're proud of

We successfully finished our agent architecture despite its complex design. Learning the ADK with no previous experience with such tools was daunting, but with enough effort, we proved it was manageable.

What we learned

A workflow of AI agents was a really alien concept to all of us, but its applications became apparent as we started to build the agents with the mission of generating in-game content through JSON structures.

What's next for EcoQuest

To finish our backend API to fully integrate every agent within the game, and release what EcoQuest was really meant to be.

Built With

• gemini-api
• google-adk
• next.js
• phaser.js
• python
• tailwind-css