SurveyBuddy

Inspiration

Over 40% of businesses now use automated calling systems. It is clear that the days of human customer service is coming to an end but it doesn't have be this way.
Often on the phone with new customer service calls, I noticed the same thing, a tacky bot voice with no empathy. That is why I decided to create SurveyBuddy, a human like survey call system.

What it does

SurveyBuddy transforms phone surveys into empathetic, AI-driven dialogues that boost response rates and depth of insights. It lets customers talk to an AI-bot system with a human voice. Users build custom surveys through an intuitive web app, then launch automated outbound calls to participants. The AI engages in real-time conversation and responds with contextual follow-ups.

How I built it

React Vite Tailwind: front end
OpenAi : transcription and context empathy
Mongo : database and querying
ElevenLabs : TTS (Text-to-speech) with human voice
TwilioDB : Call flow
Whisper: Voice Transcribing

Challenges we ran into

Correct error management: Due to complexity of the multitude of API calls, applications, and webhooks, fixing the correct component took longer than expected
Deprecated API: Extensive errors handling and deciding whether issues were a software or application issue

Accomplishments that we're proud of

End-to-End Prototype: Delivered a working system where users can create a survey, trigger a phone call, and receive transcribed, empathetic interactions with downloadable results
Robust Data Handline:Implemented secure MongoDB storage and JSON exports, enabling easy integration with analysis tools like Excel or Python scripts

What I learned

API orchestration: Dealing with so many application that depended on each other taught me about faults of latency voice processing applications and its latency
Efficiency: This being my 2nd time at HackUTD taught me to better allocate my time and make a more well rounded project

What's next for SurveyBuddy

Analytics and Insights:Build a dashboard with visualizations (e.g., sentiment analysis from responses) using libraries like Chart.js or integrate with tools like Google Analytics.
Exports and Integrations: Support CSV/Excel formats and more APIs for better workflows
More customizability: Making UI easier for administrator, selectable voice options and mass call options