Inspiration
I wanted to create something beyond a static dashboard — a system where AI isn’t just answering questions, but shaping how people interact with information. The inspiration came from seeing how teams juggle data dashboards, creative tools, and external APIs. Runi was built from the idea of turning those pieces into smart cards that an AI agent can generate, adapt, and share across multi-user sessions in real time.
What it does
Runi is a multi-user, agentic AI dashboard where information and creativity come alive through interactive cards. Cards can pull from APIs, sync through Firebase RTDB, or generate knowledge instantly via sources like WikiAPI. Beyond business use cases, Runi also includes creative tools like an AI-powered image generator and a piano keyboard that records music. Users can customize their workspace with features like dynamic backgrounds and a glassmorphism-style start menu to quickly access cards and apps.
How we built it
Frontend: Vite + React for dynamic card rendering, including both React-based cards and sandboxed iframe cards.
Backend: Firebase (Firestore, RTDB, and Storage) for session state, files, and memory indexing.
AI Agent: Runi decides which type of card to generate, manages workflows, and integrates external APIs.
Collaboration: RTDB enables real-time, multi-user sessions where updates and card changes are instantly visible to everyone.
Creative Extras: Custom-built components like the piano keyboard and background changer make Runi both useful and fun.
Challenges we ran into
Designing an AI agent that can choose the right card type automatically (React vs. iframe).
Managing real-time collaboration without data conflicts across multi-user sessions.
Balancing openness with security when connecting cards to APIs.
Creating a system that works equally well for business dashboards and playful, creative exploration.
Accomplishments that we're proud of
Building a working prototype where multi-user sessions sync in real time.
Giving Runi the ability to generate knowledge cards directly from WikiAPI.
Expanding the system beyond business with creative tools like music recording and AI image generation.
Designing a UI that feels smooth and modern, with a customizable start menu and background.
What we learned
We learned how to integrate multiple Firebase services into one cohesive system, how to let an AI agent dynamically choose different card renderers, and how to balance UX with technical complexity. We also saw the value of blending serious productivity tools with playful creative ones to make the platform more engaging.
What's next for Runi: Multi-User Agentic AI Dashboard
Next, we want to expand Runi’s capabilities with version control for projects, deeper real-time collaboration features, and more extensibility so users can build and share their own custom cards. We’re also exploring ways for Runi to self-improve by analyzing console logs from iframe cards, recursively updating itself to handle more complex problems.
Log in or sign up for Devpost to join the conversation.