DEV Community

maajid Jan
maajid Jan

Posted on

Building an Autonomous Multi-Agent Video Production Co-Pilot with Google Gemini 3.5 Flash

Disclaimer: I created this piece of content for the purpose of entering the Google All Things Agentic Hackathon on Devpost.


🚀 The Inspiration: The Hardware Creator's Dilemma

Every day, students, roboticists, and makers build incredible electronic devices—from Arduino ultrasonic radars to autonomous obstacle-avoiding rovers. However, sharing these creations with the world is challenging. Converting silent, raw workshop footage into an engaging, kid-friendly educational video tutorial requires hours of tedious work:

  • Identifying and breaking down complex electronic components into intuitive analogies.
  • Writing a paced script synchronized to circuit events.
  • Creating visual overlays and reaction graphics.
  • Designing sound effects (SFX) to emphasize key circuit beats.

For the Google "All Things Agentic" Hackathon (Taskmaster Track), I built Autonomous Agentic Maker Studio—an intelligent multi-agent AI co-pilot that transforms raw hardware footage into complete, broadcast-ready tutorial packages.


🏛️ System Architecture

Rather than relying on a single monolithic prompt, the system is designed as a decoupled multi-agent state machine powered by Google Gemini 3.5 Flash and the Google GenAI SDK:

Top comments (0)