Inspiration

We wanted to make it easier for students to get started on their job search. Filling out forms can feel like a chore—especially when you’re already stressed about graduating and finding a job. So we thought: what if you could just talk to an assistant that helps you get registered, hands-free and hassle-free?

What it does

FindMyBun is a voice-powered assistant that helps students sign up for a career platform by having a simple conversation. It asks a few questions, listens to your responses, and collects key details like your name, school, graduation year, and career goals—then stores it all for future use. No forms, just talking.

How we built it

We used OpenAI’s real-time speech-to-text to transcribe the student’s voice, then ran that through an agent built with OpenAI’s new Agents SDK to extract key info. We used ElevenLabs to generate a friendly voice response and stored everything in ChromaDB so it’s searchable later. It’s like talking to a really smart, helpful friend.

Challenges we ran into

  1. Getting real-time voice input and output to work smoothly
  2. Making the assistant understand natural speech and still pull out structured info
  3. Keeping the conversation flowing without it feeling robotic
  4. Stitching together multiple APIs under time pressure

Accomplishments that we're proud of

  1. Built a fully working voice onboarding experience in less than two days
  2. Got multiple AI tools to play nicely together in real time

What we learned

  1. How to build real-time voice pipelines with AI
  2. Ways to guide open-ended conversations and still collect accurate data
  3. How important small details (like pauses or confirmations) are for making voice feel natural
  4. That good tech doesn’t have to be complex—it just needs to be helpful

What's next for FindMyBun We want to turn this into a full onboarding flow for career services—maybe even add resume help or job matching later on. We're also excited about working with schools or job platforms to test it out with real students and keep improving based on feedback.

Built With

Share this project:

Updates