Skip to main content

About the project

(Presentation in the Github)

Imagine a robot working in the real world, with no cloud connection and limited power and hardware. A big AI model is smart but heavy and slow. A small model is fast and efficient but can only handle simple tasks.

So we use both. The robot runs two local agents that communicate. The small one handles simple tasks, and when it can’t solve something, it passes the task to the big one.

This saves energy, speeds up responses, and still gives the robot strong reasoning when it needs it. Today we’ll show how the two agents share the work

Built with

pythongeminiollamaclaudegemmahuggingface

AI tools

Coded withClaude CodeGitHub Copilot

Prize Categories Entered

  • Deepmind Google

    Best Use of Gemma 4

    Submitted
  • DigitalOcean

    Best Open-Source AI Project

    Submitted
  • CCod3rs

    Score the most points in Hack The Flag

    Submitted