OnCue

Inspiration

Anyone who's organized a large event knows how chaotic things can get behind the scenes. Volunteers need help, equipment breaks, schedules change, and small issues can quickly turn into bigger problems.

We wanted to make managing events easier for the people who are constantly running around making sure everything works. That's where OnCue comes in.

What it does

OnCue is an AI-powered event logistics assistant that combines Meta smart glasses, a mobile app, and Google's Gemini AI.

The idea is to let event staff report problems hands-free, without constantly needing to pull out their phones.

Our AI agent analyzes reports, determines urgency, recommends actions, and remembers previous information about an incident. It can also manage multiple incidents independently, helping organizers stay on top of everything happening throughout an event.

How we built it

We built OnCue using React Native and Expo for the mobile app, TypeScript and Hono for the backend, and Google's Gemini API for the AI logistics agent.

We use Zod to validate the agent's structured responses, making sure it consistently returns useful information like urgency, issue categories, and recommended actions.

We also developed a conversation and incident management system so Gemini can understand follow-up reports without mixing up different issues.

For the smart glasses integration, we're working with Meta's Wearables Device Access Toolkit.

Challenges we ran into

One of our biggest challenges was getting all the different technologies to work together, especially setting up iOS development and navigating the requirements for Meta smart glasses.

We also ran into problems with Gemini model availability, API reliability, and getting the AI to consistently return structured responses.

Another challenge was making sure everyone could work on their own part of the project without constantly waiting for another system to be finished.

Accomplishments that we're proud of

We're proud of getting our Gemini logistics agent working with realistic event scenarios.

We tested situations involving broken equipment, supply shortages, missing speakers, medical emergencies, and volunteer shortages.

The agent can understand when an issue becomes more urgent, remember follow-up information, and manage multiple incidents without confusing them.

Seeing the AI understand an evolving situation instead of just responding to individual messages was one of our biggest milestones.

What we learned

Building OnCue taught us a lot about what it takes to turn AI into something genuinely useful in a real-world environment. We learned that it's not enough for an AI to understand a problem; it also needs context, access to relevant information, and the ability to help people take action.

One of our biggest takeaways was understanding how quickly situations can change during live events. Something as simple as a broken projector might not seem urgent until you realize a presentation starts in three minutes. Teaching our AI to recognize these changes and respond accordingly was a huge part of our development process.

We also learned a lot about integrating different technologies, including Meta smart glasses, Google's Gemini AI, mobile development, and backend systems. Making these components work together showed us the importance of designing a system where information can move quickly and reliably.

Most importantly, we learned how much potential AI has to improve event management. Rather than replacing event organizers, OnCue is designed to give them another tool to stay informed, make decisions faster, and focus on creating a better experience for everyone attending.

What's next for OnCue

We see OnCue becoming a complete event logistics platform that changes how organizers manage large-scale events.

Our vision is to make OnCue an AI-powered assistant that works alongside event staff throughout an entire event. With Meta smart glasses, organizers can communicate hands-free, report issues as they happen, and receive important updates without constantly checking their phones.

We want OnCue to become more proactive, using live event schedules, staffing information, inventory, and ongoing incident reports to identify potential problems before they become major disruptions. Instead of waiting for someone to notice an issue, the system could recognize when resources are running low, when an area is understaffed, or when an upcoming session might be affected by an unresolved problem.

We also envision a centralized dashboard where event leads can monitor operations in real time, track tasks, manage volunteers, and see which issues need immediate attention. The AI would help coordinate responses, recommend the right people for specific tasks, and keep everyone updated as situations change.

Eventually, we'd love to see OnCue used beyond hackathons, including conferences, concerts, festivals, conventions, and other large-scale events.

Our ultimate goal is for OnCue to feel like an additional member of the logistics team: always aware of what's happening, ready to help, and making sure everyone is on cue.