About AURA Studio

AURA Studio is a real-time creative copilot for founders and marketers. You describe a product or campaign idea (by voice or text) and AURA streams back a structured campaign plan while also generating visual “Hero Image” assets using Gemini’s interleaved text + image output.

Inspiration

I wanted an assistant that feels like a live creative partner—not a “submit and wait” chatbot. The goal was to make ideation fast, visual, and actionable, especially for hackathon demos and real product launches.

What it does

  • Streams campaign strategy and creative direction in real time (SSE streaming)
  • Accepts voice input (Speech-to-Text) or typed briefs
  • Produces interleaved images during generation and automatically collects them in an Assets panel
  • Provides a polished “studio” UI with a dark glassmorphism aesthetic

How I built it

  • Frontend: React + Vite + Tailwind for the Studio UI
  • Backend: Node.js + Express API with Server-Sent Events streaming
  • AI: Gemini via Google’s GenAI SDK with responseModalities: [TEXT, IMAGE]
  • Speech: Google Cloud Speech-to-Text for voice transcription
  • Deployed as a single full-stack service on Google Cloud Run, built with Cloud Build and stored in Artifact Registry

Challenges

  • Getting reliable streaming UX while updating UI incrementally
  • Handling Gemini interleaved image chunks and presenting them cleanly as downloadable assets
  • Cloud Run + Cloud Build deployment troubleshooting (IAM, build config, and production startup issues)

What I learned

I learned a lot about real-time streaming patterns (SSE), production deployment on Cloud Run, and how to build a UI that makes AI output feel “alive” and easy to use.

Built With

Share this project:

Updates